Distribution of structural funds : equity , effi ciency and public procurement ( Czech Republic )

Th is article deals with structural funds allocation for projects realized by Czech regional and local authorities in the programming period 2007-2013. Geographical distribution of structural funds in the context of the equity and effi ciency goals is mapped. Two research questions are formulated in this regard. Th e fi rst research question concerns the position of lagging regions in structural funds allocation. Th e second research question maps spatial spillovers of public procurement contracts supported by structural funds. Th e fi ndings point at a rather disadvantaged position of lagging regions in structural funds allocation. Th is position is further worsened by spatial spillovers of public procurement contracts. Th us, spatial coherence between the EU and Czech regional policies is questioned.


INTRODUCTION
Public fi nance is a considerable topic of economic research. One strand of research deals with the relationship between public fi nance and economic growth. In this regard, the position of public fi nance in economic growth theories (see, e.g., Auteri and Costantini, 2004;Barro and Sala-i-Martin, 1992) and the impact of public fi nance on economic growth (see, e.g., Afonso and Furceri, 2010) are discussed. Fiscal decentralization adds the spatial dimension into the discussion (see, e.g., Rodríguez-Pose and Ezcurra, 2011;Davoodi and Zou, 1998).
Th e other strand of research is positioned in the interplay between the equity and effi ciency goals of public fi nance allocation. From the spatial perspective, the equity goal is connected with public fi nance allocation in lagging regions, while the effi ciency goal with public fi nance allocation in main growth regions (see, e.g., Morgenroth, 2010). Note an increasing political emphasis on the spatial allocation of public funding. Th is is especially true for the EU territory because territorial cohesion was inserted into the Lisbon Treaty alongside economic and social cohesion (see, e.g., Camagni, 2009;Golobic and Marot, 2011;Novosák et al., 2013). It is highly desirable, therefore, to map the geographical pattern of public fi nance allocation.
Despite its importance, mapping of geographical distribution of public fi nance has several limitations. Firstly, Blažek and Macešková (2010), Morgenroth (2010), Cabras (2011), Heald and Short (2002) point out the lack of spatially disaggregated data. Secondly, Blažek and Macešková (2010), Dall´erba and Le Gallo (2008) speak about diff erent geographical distribution of public fi nance when considering real allocation on one hand and benefi ciaries' location (headquarter) on the other. Th irdly, Heald and Short (2002) note limited knowledge about spatial spillovers of public funds. Th us, for example, geographical distribution of public fi nance may be changed by the allocation of public procurement contracts (see, e.g., Cabras, 2011;Peck and Cabras, 2011). Altogether, geographical distribution of public fi nance remains often unknown.
Th is paper is embedded in the aforementioned theoretical framework, including the three research limitations. Th e goal is to map geographical distribution of EU structural funds (hereafter referred to as SFs) in the context of the equity and effi ciency goals. Moreover, the role of public procurement contracts is evaluated. Th e focus of this paper is on projects supported by SFs in the programming period 2007-2013 (excluding the European territorial cooperation goal) and realized by regional and local authorities in the Czech Republic. Note that SFs were a key instrument of, and regional and local authorities were important stakeholders of Czech regional development strategies in this period (see, e.g., Wokoun, 2007). Th e main results of this paper do not indicate higher SFs allocation in lagging regions (see, e.g., Hájek et al., 2012;Hájek, et al., 2014 for similar conclusions). Th e position of lagging regions was, in addition, worsened by the allocation of public procurement contracts.
Th e article is structured as follows. Th e next section summarizes the main fi ndings from literature review. In section 2, the research methodology is introduced. Th e third and the fourth sections discuss the main fi ndings from the research. Th e last section concludes.

LITERATURE REVIEW
Th e research on geographical pattern of public fi nance is closely related to fi scal transfers. Th e essence of fi scal transfers is to redistribute public fi nance from relatively "rich" to relatively "poor" territorial units. To create and secure uniform living standards through the territory is the main goal of fi scal transfers in Germany, for example (see, e.g., Jüßen, 2006;Hepp and von Hagen, 2011). Th e arguments of economic ineffi ciency and political instability arising from large territorial disparities are used to substantiate fi scal transfers (see, e.g., Boldrin and Canova, 2001;Heidenreich and Wunder, 2008). However, real allocation of public fi nance may be diff erent from the "rich-poor" gradient. At least two explanations may be given: -Firstly, Hepp andvon Hagen (2011), Schraff (2014), Dellmuth and Stoff el (2012) point out asymmetries of negotiating power of particular territorial units due to, among others, their level of economic development and number of inhabitants. Lagging regions are often disadvantaged in this regard (see, e.g., Rodríguez-Pose and Ezcurra, 2011). -Secondly, Jaliu and Radulescu (2013), Kaufmann and Wagner (2005), Cappelen et al., (2003), Novosák et al., (2015), Wildowicz-Giegiel & Wyszkowski (2015) mention asymmetries in absorption capacity of particular territorial units. Lagging regions are often disadvantaged in this regard because of their low capability to prepare and submit good projects. Th is could be notably related to some themes, such as business environment (see, e.g., Belás et al., 2014;Belás et al., 2015) and innovations (see, e.g., Hájek, Grebeníček and Novosák, 2011;Kaufmann and Wagner, 2005). Th e aforementioned considerations are highly relevant also for SFs and their contribution to the equity and effi ciency goals. In this respect, a number of studies dealt with the geographical distribution of SFs. Lolos (2009), Dall'erba andLe Gallo (2008), Esposti and Bussoletti (2008), Beugelsdijk and Eijffi nger (2005) evaluated the impact of SFs allocation on regional convergence. Novosák et al., (2015), Crescenzi (2009), Crescenzi, De Fillipis andPierangeli (2015), Schraff (2014), Lambrinidis, Psycharis and Rovolis (2005), Dellmuth and Stoff el (2012), Hájek et al., (2012), Smékalová et al., (2015) searched for spatial determinants of SFs allocation. Th e fi ndings of these studies are rather ambivalent with respect to the equity and effi ciency goals, depending on the spatial level and time (ex-ante, ex-post) of evaluation. Ex-ante evaluation at the NUTS 2 level points at higher SFs allocation in lagging regions; however, as expected because of the eligibility criteria for inclusion of regions into the convergence objective. Ex-post evaluation at lower spatial levels provides less obvious results. Lolos (2009) speaks about higher SFs allocation in lagging regions in Greece while Morgenroth (2010) claims the opposite for Irish regions. Nevertheless, this research is still scarce because of the lack of spatially disaggregated data (see, e.g., Heald and Short, 2002;Blažek and Macešková, 2010;Morgenroth, 2010;Cabras, 2011).
Mapping of the geographical distribution of SFs has two additional limitations. Firstly, Hájek et al., (2012) showed a diff erent geographical distribution of SFs for two types of evaluation: (1) real SFs allocation and (2) SFs allocation based on benefi ciaries' location (see, e.g., Dall´erba and Le Gallo, 2008;Dupont and Martin, 2006 for this problem). Secondly, Heald and Short (2002) note that real SFs allocation is infl uenced by public procurement contracts. Th us, SFs allocation in one territorial unit may be squeezed out elsewhere in this way (see, e.g., Cabras, 2011;Peck and Cabras, 2011). Cabras (2011), Peck and Cabras (2011) claim that lagging regions are disadvantaged especially in competitive tendering for specialized goods and services due to their low absorption capacity in these thematic areas.
Th is paper integrates the research design as follows. Firstly, geographical distribution of SFs is mapped, considering projects realized by regional and local authorities. Th e importance of regional and local authorities for both, regional and local development and public procuring is respected in this approach. Th e equity and effi ciency goals are analyzed for the both types of evaluation: (1) real SFs allocation and (2) SFs allocation based on benefi ciaries' location. Secondly, geographical distribution of public procurement contracts, contracted by regional and local authorities and supported by SFs is mapped. Spatial spillovers from and to lagging regions are quantifi ed in this way.

METHODOLOGY
To solve the problem of the lack of spatially disaggregated data, the methodology of this paper was based on particular projects and public procurement contracts. Two databases were, therefore, compiled. Th e fi rst database contained projects realized by regional and local authorities and supported by SFs in the programming period 2007-2013 (excluding the European territorial cooperation goal). Th e Regional Information Service of the Center for Regional Development of the Czech Republic (hereafter referred to as RIS CRD) was the source of data. Th e state in January 2013 was evaluated. Th e second database contained public procurement contracts contracted by regional and local authorities and supported by SFs which were listed in the Czech offi cial Information System of Public Contracts (hereafter referred to as IS PC) in January 2013. Furthermore, several attributes of projects and public procurement contracts were added into the databases (see table 1). Th ese included also two spatial attributes for defi ning lagging regions (see table  1 for their description): -Firstly, the defi nition of the so called regions with concentrated state support (hereafter referred to as RCSS) was used. Lagging regions were, therefore, understood in accord with Czech regional policy. -Secondly, the importance of regional hierarchy was emphasized for defi ning lagging regions. Th ese are understood as peripheral areas of the Czech Republic. Table 1 Attributes of projects and public procurement contracts

Attributes Description and information source Values
Regional policy -Projects categorized according to the place of their realization or benefi ciaries' location -Public procurement contracts categorized according to benefi ciaries' location -The list of lagging regions defi ned using the offi cial Government resolutions at the level of the administrative districts of municipalities with extended competence -RIS CRD for projects; IS PC for public procurement contracts 1. Lagging regions -Structurally affected regions -Economically weak regions -Regions with high unemployment 2. Non-lagging regions 3. Prague Regional hierarchy -Projects categorized according to the place of their realization or benefi ciaries' location -Public procurement contracts categorized according to benefi ciaries' location -The levels of regional hierarchy defi ned as quartiles of population density at the level of the administrative districts of municipalities with extended competence -RIS CRD for projects; IS PC for public procurement contracts 1. The highest level of regional hierarchy 2. High level of regional hierarchy 3. Low level of regional hierarchy 4. The lowest level of regional hierarchy 5. Prague SFs allocation -SFs allocation per project/public procurement contract and place of realization/ benefi ciary's location -RIS CRD for projects; IS PC for public procurement contracts

SFs allocation in CZK
Note: In a number of cases, more places of project realization were given in the RIS CRD. However, the information about SFs allocation was at the project level only. Th erefore, SFs allocation was distributed between the places as follows. If the benefi ciary's location was diff erent from all the places of project realization SFs allocation was distributed evenly between these places. If the benefi ciary's location was the same as any place of project realization four times higher SFs allocation was assigned to this place.
Source: own compilation.
O. Hájek, J. Novosák, J. Nekolová, L. Smékalová Distribution of structural funds: equity, effi ciency and public procurement (Czech Republic) Th e compiled databases were evaluated in two steps. Firstly, geographical distribution of SFs was visualized on maps at the level of the administrative districts of municipalities with extended competence. Note that SFs allocation was normalized per one inhabitant to control for diff erent population size of particular territorial units. Th e fi rst insight into the position of lagging regions was provided in this way. Subsequently, standard methods of descriptive and inferential statistics were used. Notably, SFs allocation per one inhabitant was calculated for the categories of the two spatial attributes -regional policy and regional hierarchy. Th e aggregate SFs allocation refers to the SFs allocation per one inhabitant of the whole category. Th e average SFs allocation refers to the mean value of the administrative districts of municipalities with extended competence belonging to the same category. In addition, the parametric t-test for two independent samples and one-way ANOVA for more than two independent samples were used to test the hypothesis that the means of the categories, excluding Prague, of the two spatial attributes are not signifi cantly diff erent. Th e traditional assumptions of the two-sample independent t-test and one-way ANOVA -approximately normally distributed dependent variable for each category and homogeneity of variances -were checked by Kolmogorov-Smirnoff and Levene's tests respectively. Th e results did not indicate a violation of these assumptions.
Geographical distribution of public procurement contracts supported by SFs was evaluated in the second step of the methodology. In this respect, a new variable -spatial closeness of public procurement contracts -was defi ned as the share of public procurement contracts that were contracted to a contractor from the same territorial unit (contractor's headquarter) as the procurer (regional and local authorities). Subsequently, spatial closeness of public procurement contracts was evaluated using the same methodological approach as in the fi rst step of the evaluation -cartographic visualisation and standard methods of descriptive (aggregate and average values of spatial closeness of public procurement contracts) and inferential (two-sample independent t-test and one-way ANOVA) statistics. Once again, the traditional assumptions of the two-sample independent t-test and one-way ANOVA were verifi ed.

EMPIRICAL RESULTS  GEOGRAPHICAL DISTRIBUTION OF SFS
Th e evaluation of SFs allocation was based on 15,634 projects realized by regional and local authorities. Th e total SFs allocation for these projects was CZK 152.2 billion. Figure 1 provides the fi rst insight into the position of lagging regions: Th e factor of regional policy does not indicate higher SFs allocation in lagging regions for the both types of evaluation: (1) real SFs allocation and (2) SFs allocation based on benefi ciaries' location. Th us, lagging regions are categorized in diff erent classes of SFs allocation. In other words, some lagging regions received very high SFs allocation while other lagging regions very low SFs allocation.
Th e factor of regional hierarchy seems to be relevant for explaining the SFs allocation on the basis of benefi ciaries' location. Th us, administrative districts of large and densely populated Czech cities received a relatively higher SFs allocation. Note that a number of regional and local authorities have their headquarters just in these cities. Th e importance of regional hierarchy decreases in the evaluation of the real SFs allocation.
Th e relatively weak position of the capital city of Prague in the both types of evaluation is noteworthy. Hájek et al., (2012) provided a diff erent conclusion when evaluating SFs allocation for all types of recipients. Consequently, regional and local authorities are crucial to weaken the position of the capital city of Prague in SFs allocation. Note: Th e break classes of the colour-scale calculated at the points of arithmetic mean less standard deviation, arithmetic mean, and arithmetic mean plus standard deviation

Figure 1. Geographical distribution of SFs -SFs allocation per one inhabitant, administrative districts of municipalities with extended competence; real SFs allocation -the map above, SFs allocation based on benefi ciaries' location (headquarters) -the map below
Source: own compilation based on the data from the RIS CRD and Czech Statistical Offi ce.
O. Hájek, J. Novosák, J. Nekolová, L. Smékalová Distribution of structural funds: equity, effi ciency and public procurement (Czech Republic) Tables 2 and 3 confi rm and extend the relevance of these fi ndings. Firstly, aggregate SFs allocation per one inhabitant is slightly lower in lagging regions understood as RCSS for the both types of evaluation. Secondly, real SFs allocation indicates even geographical distribution of SFs for the categories of the both variables -regional policy and regional hierarchy. Th irdly, the evaluation based on benefi ciaries' location indicates low SFs allocation in economically weak regions and high SFs allocation in densely populated regions. Overall, the results do not support the assumption of higher SFs allocation in lagging regions because of (1) low coherence between the spatial focus of EU Cohesion Policy and Czech regional policy (see, e.g., Hájek et al., 2014); (2) low absorption capacity of lagging regions (see, e.g., Jaliu and Radulescu, 2013); and (3) higher concentration of regional and municipal authorities in densely populated regions. Th us, the strategy of "pouring sugar over cake" seems to be relevant for the evaluation. However, such a strategy does not prefer SFs allocation in lagging regions to SFs allocation in non-lagging regions. Table 2 Geographical distribution of SFs -regional policy Source: own compilation based on the data from the RIS CRD and Czech Statistical Offi ce Table 3 Geographical distribution of SFs -regional hierarchy Th e aforementioned fi ndings indicate balanced SFs allocation between lagging and non-lagging regions. Question remains whether the diff erences in the mean values of SFs allocation for the defi ned categories of regional policy and regional hierarchy, excluding the capital city of Prague, are statistically sig-nifi cant. Th is question was tested using suitable tests for continuous dependent variable. After checking for their traditional assumptions (approximately normally distributed dependent variable for each category and homogeneity of variances), the parametric t-test for two independent samples and one-way ANOVA were applied. Table 4 reports the results as signifi cance values (p-values) indicating that the diff erences in the mean values of SFs allocation are not statistically signifi cant for either of the variables, regional policy and regional hierarchy. Th us, also statistical tests do not support the assumption of higher SFs allocation in lagging regions. Th ese results have also several implications. Firstly, the impact of SFs allocation on the equity goal of Czech regional policy is unclear. Secondly, territorially based instruments may increase the coherence between SFs allocation and Czech regional policy. Th irdly, territorial impact assessment (TIA) is highly desirable in order to recognize the spatial impact of SFs allocation. Table 4 Signifi cance values (p-values) -tests of the hypotheses that the means of SFs allocation for the defi ned categories are not signifi cantly diff erent (regional policy, regional hierarchy)

Attributes
Signifi cance value Real allocation Benefi ciaries' location Regional policy -two categories 0.596 0.387 Regional policy -four categories 0.875 0.289 Regional hierarchy 0.198 0.144 Note: Th e parametric t-test for two independent samples and Levene's test for equality of variances for the fi rst case; the one-way ANOVA and Levene's test for equality of variances for the remaining two cases Note: Excluding the category "Prague" Source: own compilation based on the data from the RIS CRD and Czech Statistical Offi ce.

EMPIRICAL RESULTS  SPATIAL CLOSENESS OF PUBLIC PROCUREMENT CONTRACTS
Th e main results from the preceding section indicate no evidence of higher SFs allocation for projects realized by regional and local authorities in lagging regions. Moreover, geographical distribution of SFs allocation for these projects may be further modifi ed by public procurement contracts. Th is section addresses this issue by analyzing 10,602 public procurement contracts which were contracted by regional and local authorities and supported by SFs. Th e spatial closeness of public procurement contracts is evaluated.
Th e fi rst step of the evaluation maps spatial closeness of public procurement contracts at the level of the administrative districts of municipalities with extended competence (see fi gure 2). In this regard, the factor of regional policy does not appear to be of high relevance for explaining diff erentiation of the spatial closeness variable because both, lagging and non-lagging regions are categorized in diff erent classes of the variable. Furthermore, high values of the spatial closeness variable for several territorial units at the highest level of regional hierarchy may be observed. However, several territorial units at the lowest level of regional hierarchy indicate high values of the spatial closeness variables as well. Altogether, the geographical pattern of the spatial closeness variable is rather blurred. Tables 5 and 6 extend the fi ndings by providing the main results from the applied methods of descriptive and inferential statistics: Note: Th e break classes of the colour-scale calculated at the points of arithmetic mean less standard deviation, arithmetic mean, and arithmetic mean plus standard deviation

Figure 2. Spatial closeness of public procurement contracts (%) -administrative districts of municipalities with extended competence (N=10,602)
Source: own compilation based on the data from the IS PC.
Spatial closeness of public procurement contracts is higher in non-lagging regions compared with lagging regions. SFs allocation through public procurement contracts is, therefore, more frequently squeezed out from lagging than from non-lagging regions. However, the average values of the spatial closeness variable indicate more complex picture (compare with fi gure 2).
Spatial closeness of public procurement contracts is relatively high in regions at the highest level of regional hierarchy. Moreover, this gradient is observed also for average spatial closeness. However, the diff erences are rather modest and statistically not signifi cant (see table 6). Table 5 Spatial closeness -regional policy and regional hierarchy  Table 6 Signifi cance values (p-values) -tests of the hypotheses that the means of the spatial closeness variables for the defi ned categories of territorial units are not signifi cantly diff erent (regional policy, regional hierarchy)

Attributes
Regional policy two categories Regional policy four categories Regional hierarchy Signifi cance value 0.376 0.516 0.437 Note: Th e parametric t-test for two independent samples and Levene's test for equality of variances for the fi rst case; the one-way ANOVA and Levene's test for equality of variances for the remaining two cases Source: own compilation based on the data from the IS PC.
Spatial fl ows of public procurement contracts between the particular categories of the two spatial attributes -regional policy and regional hierarchy -are shown in tables 7 and 8. Th e weak position of lagging regions and regions at the lowest level of regional hierarchy is noteworthy. Th us, high shares of public procurement contracts are squeezed out from these two types of regions and, simultaneously, low shares of public procurement contracts fl ow into these two types of regions. Note even weaker position of lagging regions and regions at the lowest level of regional hierarchy for the fi nancial evaluation. Altogether, the weak position of lagging regions is caused by their low capability to absorb some types of public procurement contracts and by their disadvantaged position in competition with other applicants from core regions. Table 7 Spatial fl ows of public procurement contracts from and to the defi ned categories of territorial units -regional policy  Table 8 Spatial fl ows of public procurement contracts from and to the defi ned categories of territorial unitsregional hierarchy From/to Regional hierarchy Share in the number of public procurement contracts Share in fi nancial allocation for public procurement contracts I  II  III  IV  V  I  II  III  IV  V   1  2  3  4  5  6  7  8  9  1  Note: I -the highest level of regional hierarchy, II -high level of regional hierarchy, III -low level of regional hierarchy, IV -the lowest level of regional hierarchy, V -Prague Source: own compilation based on the data from the IS PC.
Th e main fi ndings from the both empirical sections do not support the assumption of higher SFs allocation in lagging regions. Th us, lagging regions did not receive a higher SFs allocation for projects realized by regional and local authorities. Moreover, lagging regions are more prone to lose higher share of SFs allocation through public procurement contracts. Consequently, these fi ndings ought to be considered in thinking about the relationship between SFs allocation and Czech regional policy.

CONCLUSIONS
Th is paper dealt with the SFs allocation for projects realized by Czech regional and local authorities in the programming period 2007-2013. Geographical distribution of SFs in the context of the equity and effi ciency goals was mapped. A special attention was paid to the position of lagging regions. Th e research consisted of two parts. Firstly, geographical distribution of SFs was evaluated. Secondly, spatial spillovers of SFs allocation through public procurement contracts were analyzed. In this way, this paper provides extended knowledge about SFs allocation than other studies (see, e.g., Lolos, 2009;Blažek and Macešková, 2010;Crescenzi, 2009).
Th e main assumption of this paper was that lagging regions received higher SFs allocation due to the goals of Czech regional policy. However, the main fi ndings of this paper did not support this assumption. Small and statistically insignifi cant diff erences were observed between lagging and non-lagging regions. In addition, lagging regions were further disadvantaged by the spatial allocation of public procurement contracts. Th us, high share of these contracts was squeezed out from lagging regions. Infl ow of public procurement contracts did not compensate for the outfl ow. Generally, low coherence between the spatial focus of EU Cohesion Policy and Czech regional policy (see, e.g., Hájek et al.,, 2014;Hudson, 2007) and low absorption capacity of lagging regions (see, e.g., Jaliu and Radulescu, 2013) may be regarded as important determinants of the results.
Several political implications may be formulated from the study. Firstly, more nuanced use of territorially based instruments may increase the coherence between SFs allocation and the equity goal of Czech regional policy. Secondly, the methodology of this paper may be of relevance for Territorial Impact Assessment (TIA) methodologies (see, e.g., Camagni, 2009;Blažek and Macešková, 2010). Th irdly, the thinking about the spatial impact of SFs allocation ought to consider various methodological issues (e.g., type of evaluation, spatial spillovers from public procurement contracts and others). Finally, limitations of the research should be considered as well. Th ese include: (1) the focus of the paper on regional and local authorities only; (2) missing thematic decomposition of SFs allocation; (3) limited research on determinants of SFs allocation; (4) missing discussion of the most suitable spatial level for evaluation. Th ese limitations may be regarded also as the future directions of research.