Asymmetric effects of EU cohesion policy on EU regional growth: The role of macroeconomic uncertainty

The Journal of Economic Asymmetries 30 (2024) e00382 Available online 27 August 2024 1703-4949/© 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Asymmetric effects of EU cohesion policy on EU regional growth: The role of macroeconomic uncertainty Mehmet Pinar a , b , * , Burhan Can Karahasan c a Departamento de An´ alisis Econ´ omico y Economía Política, Universidad de Sevilla, Avda. Ram´ on y Cajal, 1, 41018, Sevilla, Spain b Department of Economics, Istanbul Bilgi University, Istanbul, Turkey c MEF University, Department of Economics, Maslak Ayaza˘ ga Cd. No:4, 34396 Sarıyer, Istanbul, Turkey ARTICLE INFO Keywords: EU funds Structural funds Cohesion policy Regional growth Uncertainty ABSTRACT Cohesion policy and the EU funds have been key elements for territorial integration in Europe. Evidence shows that EU funds support the growth performance of regions. However, less has been discussed about the potential impact of macroeconomic uncertainty on the effectiveness of EU funds. Our analyses confirm that EU funds are important in understanding regional economic growth differences. However, the extent of macroeconomic uncertainty decreases the effectiveness of the EU funds. Our results are robust in including local controls, non-linearity of the EU funds’effect, different EU fund categories, and regional heterogeneity in the EU. 1. Introduction The Cohesion Policy is the main investment policy of the European Union (EU), and one-third of the EU budget, roughly 392 billion Euros, is planned to be distributed under the umbrella of this policy across different regions between 2021 and 2027 (European Commission, 2023). While the Cohesion Policy addresses a different set of development goals, most of the funds are distributed to increase the economic and social integration of the EU regions and reduce social and economic disparities across its regions. The primary objective of the Cohesion Policy is to harmonize the economic and social life across the EU regions. While the Cohesion Policy covers all of the EU regions, priority is directed to those regions with a per capita income below 75% of the EU average. 1 During the different phases of enlargement, regional heterogeneity accelerated, and the policy increasingly has become a key institutional tool for spreading the benefits of integration. In other words, the Cohesion Policy is an instrument aiming for balanced development across the EU regions. While the core objective of providing a more equal regional development stands still, historical developments show the changing nature of the tools and objectives. Some of the core aims of the Cohesion policy nowadays are to foster competitiveness, employment, economic growth, and sustainable development (Brunazzo, 2016). However, from the early foundation years to today, there have been periodic changes in the priorities and implications of the policy. Begg (2010) confirms that even though regional development is the cohesion policy’s primary objective, historical evolution brings additional objectives. Therefore, tracing the main developments of the policy is essential. The need to consider the regional dimension in policymaking is an inevitable outcome of the enlargement process. Before the UK and Ireland’s accession to the EU, most of the * Corresponding author. Departamento de An´ alisis Econ´ omico y Economía Política, Universidad de Sevilla, Avda. Ram´ on y Cajal, 1, 41018, Sevilla, Spain. E-mail addresses: [email protected] (M. Pinar), [email protected] (B.C. Karahasan). 1 See https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32021D1130 for a list of regions that are eligible to receive funding under the 2021–2027 round of the Cohesion Policy (accessed on 17 July 2024). Contents lists available at ScienceDirect The Journal of Economic Asymmetries journal homepage: www.elsevier.com/locate/jeca https://doi.org/10.1016/j.jeca.2024.e00382 Received 29 May 2024; Received in revised form 29 July 2024; Accepted 21 August 2024 The Journal of Economic Asymmetries 30 (2024) e00382 2 concerns about regional policies were directed to the gap between Italy and the rest of the union. Later on, the move from the European Social Fund towards the European Regional Development Fund constitutes an important step towards the implementation of a more holistic policy set to mitigate the adverse effects of regional disparities (see Begg (2010) and Bachtler and Mendez (2020, pp. 121–139) for more details). Eislet (2008)underlines the three main objectives that are shaped through the enlargement process: (i) before the 1980s, the core objective was compensation; (ii) after 1988, the allocation objective became more central; (iii) after the 1990s, more weight is given to the productivity approach. These developments show the changing nature of the approach implemented within the Cohesion Policy. It has to be noted that historically, priorities of funds and policy change have also changed. For instance, during the early 1980s, priorities were on correcting the regional disparities due to agricultural dominance, industrial change and structural employment. During the 1990s, more discussions were directed at decentralization, followed by simplification procedures in the early 2000s. These developments are mostly followed by concerns about the asymmetric growth of population and income with the enlargement processes as the population growth exceeds the income acceleration during the historical enlargements. After the 2000s, more policy discussion is diverted to growth and job creation. Inevitably, sectoral priorities become more central to understanding the asymmetric economic progress across the EU regions. An extensive number of articles have examined the economic implications of the EU’s Cohesion policy. Most of the existing literature found positive effects of structural funds on economic growth (see e.g., Becker et al., 2010; Biedka et al., 2022;Di Cataldo & Monastiriotis, 2020;Pellegrini et al., 2013) and Cohesion funds help EU regions to converge in standard of living (e.g., Calegari et al., 2023;Caraveli et al., 2008;Kutan &Yigit, 2007;Maynou et al., 2016). From a wider perspective, Canova (2024) underlines that the positive aspects of structural funds under the Cohesion Policy are mostly visible across the EU territories in the medium and long run. However, a limited number of studies demonstrated that the effect of the funds was not significant in promoting economic growth (e.g., Psycharis et al., 2020). Recent literature also identified that the effectiveness (ineffectiveness) of the Cohesion policy funds depends on the region and country conditions. It has also been found that the funds were more effective in institutionally strong regions (Barbero et al., 2023; Rodríguez-Pose &Garcilazo, 2015) and the magnitude of funds allocated to the regions also played a role in the effectiveness of the funds (see e.g., Becker et al., 2012; Cerqua &Pellegrini, 2018;Fiaschi et al., 2018). Our paper aims to contribute to this stream of literature by providing an additional explanation of why EU structural funds may not be effective in promoting economic growth in some countries by considering the macroeconomic uncertainties in the EU countries. We highlight that among various challenges of the EU, regional economic differences and the potential convergence struggle are vital dimensions to motivate the importance of the Cohesion policy (see e.g., L´ opez-Bazo et al. (1999), Carevell i et al. (2008), Arbia et al. (2008), Rauhut and Humer (2020) for details). 2 Macroeconomic uncertainty is found to have implications for various macroeconomic indicators such as aggregate output, investment, consumption, inflation, and unemployment, among many others (Bloom, 2009). Firstly, macroeconomic uncertainty leads to lower economic activity and higher unemployment levels. Schaal (2017) demonstrates that the uncertainty during the 2007–2009 financial crises accounted for about 40% of the total increase in unemployment in the United States. On the other hand, Choi and Loungani (2015) show that while aggregate uncertainty shocks have a short-run effect on an increase in unemployment rates, but sectoral uncertainty shocks have a long-term impact on unemployment in the United States. Furthermore, existing studies showed that uncertainty has an asymmetric effect on unemployment rates and the effects of uncertainty on unemployment were found to vary across US states (Ahmed et al., 2022), during recession and boom periods (Caggiano et al., 2017) and over time (Eksi &Tas, 2022). Secondly, uncertainty has implications for economic growth and inflation. Neanidis and Savva (2013) show that higher inflation uncertainty diminishes growth rates during high inflation periods. Chowdhury et al. (2018) also demonstrate that inflation uncertainty harms output growth in the United Kingdom and the United States. Chowdhury (2024) finds that uncertainty significantly affects output growth and inflation in developing countries. Berger et al. (2017) show increased global uncertainty hampers economic growth and national inflation rates. Bianchi et al. (2023) empirically show that both demand and supply uncertainty lowers economic activity, and supply-side uncertainty increases inflation and lowers investment activities. Furthermore, due to increased risks associated with the uncertainty, the existing literature found that macroeconomic uncertainty leads to lower private investment (Aizenman &Marion, 1993;Demir, 2009;Asamoah et al., 2016;Meinen &Roehe, 2017;Adil &Roy, 2024), household consumption (Coibion et al., 2024; Ghirelli et al., 2021;Nam et al., 2021)), foreign direct investment (e.g., Canh et al., 2020;Choi et al., 2021;Choi et al., 2023;Nguyen & Lee, 2021), credit market activity (Alessandri &Bottero, 2020;Gozgor et al., 2019) and stock market returns (Batabyal &Killins, 2021; Nusair &Al-Khasawneh, 2022). Based on the existing literature, our main argument is that EU structural funds are less effective in countries with higher macroeconomic uncertainty due to decreased economic activity and increased risk. While the existing literature examined the implications of uncertainty on macroeconomic indicators (e.g., output, investment, consumption, inflation, unemployment, foreign direct investment, stock market returns, etc.) and examined the effectiveness of the Cohesion policy, there is no investigation of the effectiveness of the EU Cohesion policy in the presence of economic uncertainty and this paper aims to close this gap. The remainder of the paper is as follows. The next section provides a literature review of the effectiveness of the Cohesion policy for economic growth and convergence and the role of uncertainty in economic outcomes. Section 3provides the data and empirical methodology. Section 4provides the empirical findings, and section 5concludes and discusses policy recommendations. 2 From a non-EU perspective, interested readers can refer to Duran (2019) for an EU candidate country (Turkey) and to Holmes et al. (2014) for an advanced non-EU country (US). M. Pinar and B.C. Karahasan The Journal of Economic Asymmetries 30 (2024) e00382 3 2. Literature review Most of the existing literature demonstrated that the Cohesion policy promotes regional economic growth (see e.g., Becker et al., 2010;Biedka et al., 2022;Di Cataldo &Monastiriotis, 2020;Pellegrini et al., 2013;Fiaschi et al., 2018;Di Cataldo &Monastiriotis, 2020, among others); however, the effectiveness of the EU cohesion policy in promoting regional economic growth varies based on different regional characteristics. On the one hand, a stream of literature found that Cohesion policy is more effective in regions with lower income per capita (Mohl &Hagen, 2010;Pellegrini et al., 2013;Di Cataldo &Monastiriotis, 2020;Biedka et al., 2022). On the other hand, another stream of literature showed that Cohesion policy is more effective in socio-economically advanced regions (Crescenzi &Giua, 2016) and areas with stronger territorial capital (Fratesi &Perucca, 2019;Bachtr¨ ogler et al., 2020) and institutional quality (Barbero et al., 2023;Rodríguez-Pose &Garcilazo, 2015;Scheurer &Haase, 2018). 3 In the meantime, the conditionality of the EU fund effectiveness also receives rising interest among scholars. Vida (2021) shows that institutional quality has a strong and positive impact on the effectiveness of the EU funds to promote economic growth and development. From a different vein, Dicharry (2023) examines the absorption capacity of the EU regions and suggests that fast absorption of the EU funds reduces the effectiveness of the Cohesion Policy. In line with developments in the literature, we examine the role of country-level macroeconomic uncertainty in the effectiveness of the EU cohesion policy, as country-level uncertainty leads to decreased firm-level and aggregate economic activities, which could hamper the effectiveness of the Cohesion policy funds. The existing literature highlighted the role of uncertainty in explaining aggregate economic activity. For instance, Nguyen (2020) demonstrates that economic uncertainty increases the probability and duration of economic recession in 10 European Union countries after examining quarterly data between the second quarter of 1987 and the first quarter of 2021. Analyzing the 2011–2020 period, Wen et al. (2022) showed that uncertainty has led to a sustained decline in economic activity in Pakistan. Adil and Roy (2024) demonstrated that economic uncertainty reduced India’s domestic investment. Nam et al. (2021) demonstrated that financial and macroeconomic uncertainty reduced the consumption growth of U.S. households in the 1986–2019 period. Economic uncertainty was also found to hamper the stock market returns in Canada and G7 countries (Batabyal &Killins, 2021;Nusair &Al-Khasawneh, 2022). Furthermore, uncertainty led to reduced capital flows and investment in countries. Ogbonna et al. (2022) demonstrated that the increased uncertainty led to decreased foreign direct investment (FDI) flows to African countries. Nguyen and Lee (2021) also found that countries with high levels of uncertainty received lower FDI flows using a sample of 116 countries in the 1996–2017 period. Similarly, Choi et al. (2023) examined the role of uncertainty for capital flows to 143 recipient countries from 51 countries, and they found that an increased uncertainty led to a major and persistent decrease in bank credit and portfolio debt and equity inflows. Similarly, uncertainty also had negative implications at the firm level. Nguyen et al. (2022) examined the impact of uncertainty on default risk in 26 emerging markets in the period 1991–2019 period, and they demonstrated that the uncertainty increased firms’ default risk. Iqbal et al. (2020) demonstrate that economic uncertainty reduced the performance of US-listed non-financial firms. Similarly, Ahsan et al. (2023) also found that policy uncertainty reduced the firm performance based on the dataset of 702 non-financial European firms. The low performance of the firms may be due to various factors. Firms could lower their investment levels during uncertain times due to lower economic activity and may be constrained due to low credit availability. For instance, using panel data from 139 countries covering the 1996–2017 period, Gozgor et al. (2019) showed that higher uncertainty lowers domestic credits. Similarly, Alessandri and Bottero (2020) analyzed more than two million corporate loan applications to Italian banks and found that increased uncertainty lowers the chances of obtaining loans and increases the waiting time to be disbursed. In other words, corporate credit supply declines during uncertainty, which lowers business activities. Along the same lines, Zhang et al. (2023) analyzed 19,778 firms from 82 countries during the 2019–2022 period and found that the increased uncertainty lowered firm-level investments. While the EU cohesion fund effectiveness literature analyzed the importance of different characteristics of the regions (e.g., institutional quality, magnitude of the funds, human capital, etc.) for the effectiveness of EU funds, the role of the macroeconomic uncertainty for the effectiveness of the EU funds was not examined. Overall, uncertainty hampers aggregate macroeconomic activities and private investment and returns to private investment. Therefore, this paper aims to examine whether the returns to the EU’s main investment are affected by macroeconomic uncertainty. In the next section, we will provide the details of the data and empirical strategy. 3. Data and empirical specification Our dependent variable is the regional growth of per capita Gross Domestic Product Data (GDP). We collect the NUTS-2 level per capita GDP data from the Cambridge Econometrics’European regional database, which is accessible via the Annual Regional Database of the European Commission’s Directorate General for Regional and Urban Policy (Annual Regional Database, 2022). As highlighted in the previous section, our main variables of interest are the regional distribution of the EU funds and country-base differences in uncertainty. First, EU funds’NUTS-2 level distribution is obtained from the Historic EU payments database of the European Commission (DG Regional Policy), which covers the 1988–2018 period (Historic EU, 2022a). The database contains seven different 3 Socio-economic development levels of the regions are measured with the social filter index, which measures the structural socio-economic conditions for the development of a well-functioning region (see Rodríguez-Pose (1999) and Crescenzi and Giua (2016) for the details of the social filter index). Territorial capital encompasses various components such as human, creative, cultural, and institutional capital levels, among many other elements (see Table 1 of Morretta (2020) for the list of factors used for measuring territorial capital). M. Pinar and B.C. Karahasan The Journal of Economic Asymmetries 30 (2024) e00382 4 domains: Cohesion Fund (CF), European Agricultural Fund for Rural Development (EAFRD), European Maritime and Fisheries Fund (EMFF), European Regional Development Fund (ERDF), European Social Fund (ESF), Fund for European aid for the most deprived (FEAD) and Youth employment initiative (YEI). 4 Our baseline models use the total funds allocated to each region. 5 Fig. 1 provides the spatial distribution of the EU funds for selected years of our sample. It is clear that the core EU regions receive the lowest amount of EU funds. In line with the central theme of the Cohesion Policy, peripheral regions of the countries that became members of the EU in different enlargement processes receive the highest proportion of the EU funds. During the 1990s, most of the funds are distributed to the southern regions of the Mediterranean countries. After the Central and Eastern Europe (CEE) enlargement process, we also observed that regions of the CEE countries started to receive high amounts of funding from the EU. Second, to account for the uncertainty, we use the World Uncertainty Index (WUI) (Ahir et al., 2022). Note that our measure of uncertainty is available at the country level for the 1952–2022 period. 6 Therefore, we match the NUTS-2 level data with the country-based uncertainty indicator. This uncertainty index relies on a single source with specific economic and political coverage, thus increasing the accuracy and consistency of the index computation and allowing cross-country comparison. The index is obtained by counting the number of uncertainty-related words in the quarterly country reports of the Economist Intelligence Unit (EIU), which was then scaled by the total number of words in each respective report. Overall, a higher value in the index suggests that there has been a higher number of uncertainty-related words in the report, suggesting a higher level of uncertainty in this country. Since its first introduction, the index has been widely used to evaluate the role of uncertainty in explaining different socioeconomic concepts by the literature (see e.g., Choi et al., 2023;Gozgor, 2022;Gozgor et al., 2019;Nguyen et al., 2022;Nguyen &Lee, 2021;Ogbonna et al., 2022;Wu et al., 2022;Zhang et al., 2023). Another uncertainty index that has been used extensively by the existing literature is economic policy uncertainty (see Baker et al. (2016) for the construction of the index and Al-Thaqeb and Algharabali (2019) for a literature review on the use of economic policy uncertainty). WUI differs from economic policy uncertainty (EPU) in three ways (see Ahir et al. (2022) for a detailed discussion). While EPU relies on a large set of newspapers to construct an uncertainty index, WUI only relies on the EIU reports, which allow Fig. 1. Geographical evolution of the EU funds. Fig. 2. Historical evolution of the uncertainty index for the EU countries. Source: Authors’own calculations based on Ahir et al. (2022). 4 See the robustness sub-section 4.3 for alternative analyses where we only use individual funds (e.g. ERDF and ESF). 5 Note that we use the modeled annual expenditures to control for the EU funds received by the regions. This has been recommended and discussed in the methodology of the data set. Details can be seen via https://cohesiondata.ec.europa.eu/Other/Historic-EU-payments-regionalised-and- modelled/tc55-7ysv (last accessed on the 12.12.2022). 6 Among different versions of the WUI, we use the three quarter moving average of the WUI. M. Pinar and B.C. Karahasan The Journal of Economic Asymmetries 30 (2024) e00382 5 comparability across countries. Secondly, while the frequency of the WUI is quarterly, the EPU index provides monthly uncertainty indices. Since our analyses rely on yearly data, the frequency of the data set does not impose any limitations. Finally, EPU only provides an uncertainty index for 22 countries; WUI offers uncertainty indices for 143 countries. In particular, based on the application area of this paper, EPU only offers an uncertainty index for 7 EU countries (i.e., France, Germany, Greece, Ireland, Italy, Spain and Sweden), but WUI provides information about the uncertainty of 23 EU countries with the exceptions of Cyprus, Estonia, Luxembourg and Malta. Therefore, our analyses rely on the WUI due to a larger EU country coverage. In Fig. 2, we sketch the historical evolution of the uncertainty index for the EU countries. Overall, there is a tendency for rising uncertainty for the whole set of countries. Notably, uncertainty is not merely a developmental phenomenon. While some relatively less developed EU countries (e.g., Bulgaria and Romania) witnessed a sharp increase in uncertainty, other developed countries, such as the UK and Germany, also experienced a rapid increase in uncertainty during the last decades. Meanwhile, northern countries like Finland and Sweden witness relatively less change in their uncertainty levels. These descriptive findings confirm the uncertainty heterogeneity across the EU member states. Finally, we also collect data on a set of control variables for econometric analyses. Our control variables are employment rate, capital per labor, per capita gross value added in industrial production, population density, and share of employment with tertiary education. Data for control variables are also collected at the NUTS-2 level (ARDECO , 2022). Our overall data forms an unbalanced panel for the period of 1990–2018. Descriptive statistics for all variables are provided in Table 1. Our main objective is to assess the impact of EU funds and economic uncertainty on the distribution of per capita GDP growth across the EU regions. Moreover, we are also interested in the potential interplay between EU funds and economic uncertainty. We argue that economic uncertainty can decrease the effectiveness of EU funds. To account for these issues, we employ the following augmented panel fixed effect model: Δln (y)i,t=βΔln (y)i,t−1+δZt+θ1lnEUi,t+θ2(lnEUi,t)2+γZtln EUi,t+φXi,t+ π t+ τ i+ ε it (1) where iand trepresent region and time, respectively; yis the per capita GDP; Zis the uncertainty index; EU is the per capita EU funds; X is a set of control variables as introduced in the previous sub-section; π tand τ iare the time and cross-section fixed effects that control for the national business cycles’effects and time-invariant regional heterogeneities, respectively. We also incorporate the square of the EU to capture the potential non-linear effects of the EU funds. The main idea for incorporating the non-linearity stems from the potential asymmetric influence of the funds for the receiving countries. In other words, we are concerned about the decreasing returns to Cohesion Policy funds and include a non-linear term in our empirical strategy (see Becker et al., 2012;Cerqua &Pellegrini, 2018;Di Cataldo &Monastiriotis, 2020 for details). Moreover, to test the main hypothesis of the paper, we also include an interaction term (ZtEUi,t).Finally, ε it is the identically and independently distributed residuals. With the exception of the uncertainty index, all variables are collected at the NUTS-2 level. As highlighted in the previous subsection, we merge the country-based uncertainty index with the NUTS-2 level regional data set in our analyses. Also, note that we cluster the standard errors at the NUTS-2 level for all of the analyses carried out in the paper. In our empirical setup, we first estimate unconditional models by disregarding the other potential factors (X) that can likely influence regional differences in per capita GDP. Next, we augment our understanding and estimate conditional models by considering these additional factors. Finally, we provide results of the panel fixed effect models with only regional fixed effects (one-way) and then with region and year fixed effects (twoway). 4. Empirical results 4.1. Baseline results To investigate the impact of EU funds and uncertainty on economic growth, we start by estimating unconditional models. Our first objective is to assess the separate impact of EU funds and uncertainty on the regional distribution of economic growth across the EU regions. Second, we are centrally interested in the potential interplay between EU funds and macroeconomic uncertainty as we expect that it decreases the effectiveness of the EU funds. While doing this, we also incorporate the non-linear effects of the EU funds. We also supply results with one-way fixed effects (cross-section fixed effects) and two-way fixed effects (cross-section and time fixed effects). Results are given in Table 2. Table 1 Descriptive statistics. Obs. Mean Stdev. Min. Max. Pc funds 5939 95.473 141.938 0.001 1241.831 Growth rate of per capita GDP 5719 0.015 0.035 −0.221 0.488 Uncertainty Index 5939 0.055 0.035 0.000 0.188 Employment Rate 5834 0.438 0.074 0.272 0.749 Capital per worker (in ln) 5834 9.295 0.568 6.301 10.749 Industrial value added per capita (in ln) 5834 8.268 0.680 5.581 9.949 Population density (in ln) 5158 5.009 1.176 1.128 8.919 Tertiary Education 4203 0.245 0.090 0.051 0.622 M. Pinar and B.C. Karahasan The Journal of Economic Asymmetries 30 (2024) e00382 6 Table 2 Impact of the EU funds and macroeconomic uncertainty on regional growth (Unconditional models). (1) (2) (3) (4) (5) (6) (7) (8) Per capita GDP (t-1) −0.0778 a −0.0952 a −0.0739 a −0.0952 a −0.0725 a −0.0980 a −0.0735 a −0.0985 a (0.0065) (0.0124) (0.0068) (0.0124) (0.0070) (0.0127) (0.0069) (0.0126) PC EU funds 0.0017 a 0.0038 a 0.0016 a 0.0037 a 0.0023 a 0.0028 a 0.0031 a 0.0033 a (0.0006) (0.0007) (0.0006) (0.0007) (0.0007) (0.0009) (0.0008) (0.0009) PC EU funds_sqr. −0.0589 a 0.0003 c −0.0002 0.0003 c (0.0123) (0.0002) (0.0002) (0.0002) Uncertainty Index −0.0586 a −0.0267 c −0.0002 −0.0268 b −0.0013 0.0089 (0.0122) (0.0136) (0.0002) (0.0136) (0.0298) (0.0286) PC EU fund c Uncertainty Index −0.0152 b −0.0096 (0.0076) (0.0069) Test for joint sign. (interaction and PC EU funds) 8.34 6.83 [0.00] [0.00] Test for joint sign. (interaction and Uncertainty) 12.01 2.87 [0.00] [0.05] Observations 5719 5719 5719 5719 5719 5719 5719 5719 R-squared 0.0735 0.3207 0.0768 0.3212 0.0772 0.3221 0.0778 0.3223 Number of regions 220 220 220 220 220 220 220 220 Regional controls No No No No No No No No Year FE No Yes No Yes No Yes No Yes Cross-section FE Yes Yes Yes Yes Yes Yes Yes Yes Notes: Clustered (NUTS-2) standard errors in parentheses. a p<0.01. b p<0.05. c p<0.1. Table 3 Impact of the EU funds and macroeconomic uncertainty on regional growth (Conditional models). (1) (2) (3) (4) (5) (6) (7) (8) Per capita GDP (t-1) −0.2130 a −0.2035 a −0.2145 a −0.2049 a −0.2807 a −0.2059 a −0.2823 a −0.2070 a (0.0214) (0.0291) (0.0217) (0.0295) (0.0393) (0.0328) (0.0403) (0.0333) PC EU funds 0.0109 a 0.0057 a 0.0120 a 0.0069 a 0.0154 a 0.0126 a 0.0164 a 0.0136 a (0.0016) (0.0013) (0.0018) (0.0015) (0.0042) (0.0040) (0.0045) (0.0043) PC EU funds_sqr. −0.0016 a −0.0007 a −0.0016 a −0.0007 a −0.0025 a −0.0017 a −0.0025 a −0.0017 a (0.0002) (0.0002) (0.0002) (0.0002) (0.0005) (0.0005) (0.0005) (0.0005) Uncertainty Index −0.0155 −0.0275 b 0.0604 0.0616 c −0.0674 a −0.0340 b 0.0055 0.0364 (0.0137) (0.0127) (0.0421) (0.0361) (0.0166) (0.0144) (0.0704) (0.0583) Employment rate 0.1816 a 0.1536 a 0.1790 a 0.1509 a 0.2469 a 0.1915 a 0.2451 a 0.1908 a (0.0345) (0.0314) (0.0345) (0.0314) (0.0507) (0.0457) (0.0508) (0.0457) Capital per worker (in ln) 0.0107 b 0.0131 a 0.0099 b 0.0122 a 0.0214 a 0.0124 b 0.0206 a 0.0114 c (0.0042) (0.0044) (0.0042) (0.0045) (0.0058) (0.0057) (0.0058) (0.0059) Industrial value added per capita (in ln) 0.1125 a 0.0815 a 0.1133 a 0.0825 a 0.1270 a 0.0865 a 0.1274 a 0.0870 a (0.0108) (0.0109) (0.0109) (0.0110) (0.0162) (0.0134) (0.0164) (0.0135) Population density (in ln) −0.0198 −0.0597 a −0.0206 −0.0607 a −0.1102 b −0.0706 b −0.1129 b −0.0729 b (0.0251) (0.0216) (0.0251) (0.0216) (0.0441) (0.0342) (0.0448) (0.0347) PC EU fund c Uncertainty Index −0.0192 c −0.0225 a −0.0173 −0.0164 (0.0098) (0.0085) (0.0161) (0.0132) Tertiary Education 0.2087 a 0.1247 a 0.2079 a 0.1281 a (0.0311) (0.0347) (0.0310) (0.0351) Test for joint sign. (interaction and PC EU funds) 24.06 10.81 6.69 4.97 [0.00] [0.00] [0.00] [0.01] Test for joint sign. (interaction and Uncertainty 2.79 6.01 8.89 3.57 [0.06] [0.00] [0.00] [0.03] Observations 4676 4676 4676 4676 3534 3534 3534 3534 R-squared 0.2476 0.4141 0.2484 0.4152 0.3205 0.4583 0.3211 0.4587 Number of regions 196 196 196 196 196 196 196 196 Regional controls No No No No Yes Yes Yes Yes Year FE No Yes No Yes No Yes No Yes Cross-section FE Yes Yes Yes Yes Yes Yes Yes Yes Notes: Clustered (NUTS-2) standard errors in parentheses. a p<0.01. b p<0.05. c p<0.1. M. Pinar and B.C. Karahasan The Journal of Economic Asymmetries 30 (2024) e00382 7 Columns (1) to (4) indicate that regardless of the fixed effects specification, those regions that receive more EU funds realize higher economic growth. This confirms the prior literature on the positive role of the EU cohesion policy in stimulating regional growth and cohesion. On the other hand, country-based uncertainty has a detrimental impact on regional economic growth (Columns (3) and (4)). Regions belonging to countries with higher macroeconomic uncertainty are realizing slower economic growth, which limits their transition to the EU average. In line with the convergence framework, lagged per capita GDP negatively impacts regional growth. We also add the potential non-linear impact of the EU funds and find a negative effect, which confirms the argument that the effectiveness of the EU funds declines after a certain threshold level (Columns (5) and (6)). Finally, we add an interaction term that aims to assess the change in the impact of EU funds based on the uncertainty index of the countries (Columns (7) and (8)). Note that the interaction term is negative, albeit only significant for the one-way fixed effect model. However, EU funds and uncertainty index are jointly significant with the interaction term for both specifications. Therefore, the negative impact of the interaction term validates that rising macroeconomic uncertainty decreases the returns of the EU funds usage. Next, we estimate the conditional models and provide the results in Table 3. These models include regional controls, the nonlinearity of the EU funds and the interaction between EU funds and the uncertainty index. As human capital data can be obtained for a shorter time interval, we estimate the conditional models without (Columns (1) to (4)) and with (Columns (5) to (8)) tertiary education variable. Once again, we estimate one-way and two-way models iteratively to show the level of sensitivity for the potential impact of nationwide business cycles and the time-invariant regional heterogeneities. Regardless of the specification, our results show that EU funds positively and significantly influence the regional economic growth of the EU regions. Once again, the lagged impact of per capita GDP is in line with the prior findings and theoretical expectations. Remarkably, the square of the EU funds is also significant yet negatively signed, which supports the view of decreasing fund productivity after a certain level. On the other hand, the uncertainty index has a significant negative impact on regional growth only for the two-way fixed effect conditional models reported in columns (2), (5) and (6). However, once we also include the interaction term (Columns (3), (4) and (7), (8)), the uncertainty index becomes positive but not significant at conventional significance levels. Remarkably, the interaction term is negative, yet only significant for the models reported in columns (3) and (4). Still, joint significance test results show that EU funds and the uncertainty index are jointly significant with the interaction term for all of the augmented models. This confirms that the impact of EU funds declines for regions belonging to countries with higher uncertainty. These findings from the conditional models validate our main argument that rising macroeconomic uncertainty decreases the effectiveness of the EU funds even after controlling for a host of other factors that can influence regional growth across the EU regions. Table 4 Impact of the EU funds and macroeconomic uncertainty on regional growth (Robustness analyses for ERD Funds). (1) (2) (3) (4) (5) (6) (7) (8) Per capita GDP (t-1) −0.2211 a −0.2504 a −0.2223 a −0.2515 a −0.3535 a −0.2715 a −0.3558 a −0.2722 a (0.0146) (0.0203) (0.0148) (0.0204) (0.0231) (0.0222) (0.0233) (0.0223) PC EU funds 0.0077 a 0.0049 a 0.0088 a 0.0058 a 0.0046 a 0.0055 a 0.0059 a 0.0060 a (0.0012) (0.0010) (0.0014) (0.0011) (0.0017) (0.0015) (0.0019) (0.0016) PC EU funds_sqr. −0.0013 a −0.0005 b −0.0014 a −0.0005 b −0.0010 a −0.0007 b −0.0010 a −0.0007 b (0.0002) (0.0002) (0.0002) (0.0002) (0.0003) (0.0003) (0.0003) (0.0003) Uncertainty Index −0.0042 −0.0326 b 0.0470 0.0112 −0.0555 a −0.0394 a 0.0068 −0.0131 (0.0131) (0.0131) (0.0303) (0.0267) (0.0156) (0.0148) (0.0368) (0.0362) Employment rate 0.1897 a 0.1630 a 0.1871 a 0.1607 a 0.2373 a 0.1550 a 0.2336 a 0.1540 a (0.0334) (0.0304) (0.0333) (0.0303) (0.0431) (0.0402) (0.0428) (0.0400) Capital per worker (in ln) 0.0155 a 0.0216 a 0.0146 a 0.0209 a 0.0301 a 0.0238 a 0.0291 a 0.0232 a (0.0041) (0.0042) (0.0042) (0.0043) (0.0054) (0.0055) (0.0055) (0.0058) Industrial value added per capita (in ln) 0.1155 a 0.0896 a 0.1162 a 0.0904 a 0.1470 a 0.1015 a 0.1475 a 0.1018 a (0.0094) (0.0095) (0.0094) (0.0095) (0.0123) (0.0112) (0.0123) (0.0112) Population density (in ln) −0.0168 −0.0780 a −0.0176 −0.0790 a −0.1488 a −0.1133 a −0.1539 a −0.1154 a (0.0261) (0.0234) (0.0261) (0.0234) (0.0466) (0.0353) (0.0472) (0.0359) PC EU fund c Uncertainty Index −0.0177 b −0.0149 b −0.0206 c −0.0084 (0.0089) (0.0075) (0.0117) (0.0101) Tertiary Education 0.2735 a 0.1232 a 0.2730 a 0.1251 a (0.0258) (0.0333) (0.0259) (0.0341) Test for joint sign. (interaction and PC EU funds) 19.97 12.91 4.96 7.3 [0.00] [0.00] [0.01] [0.00] Test for joint sign. (interaction and Uncertainty 2.11 5.68 7.02 4.18 [0.12] [0.00] [0.00] [0.01] Observations 4524 4524 4524 4524 3392 3392 3392 3392 R-squared 0.2491 0.4397 0.2501 0.4403 0.3683 0.4934 0.3696 0.4936 Number of regions 196 196 196 196 196 196 196 196 Regional controls Yes Yes Yes Yes Yes Yes Yes Yes Year FE No Yes No Yes No Yes No Yes Cross-section FE Yes Yes Yes Yes Yes Yes Yes Yes Notes: Clustered (NUTS-2) standard errors in parentheses. a p<0.01. b p<0.05. c p<0.1. M. Pinar and B.C. Karahasan The Journal of Economic Asymmetries 30 (2024) e00382 8 Overall, EU funds are an important dimension of cohesion. Those regions that receive higher EU funds are realizing high per capita GDP growth even after controlling for other factors that can affect regional economic growth. On the other hand, we also find out that macroeconomic uncertainty matters. Those regions belonging to countries with higher uncertainty are realizing lower economic growth. Interestingly, once we incorporate the interaction term, the direct impact of the uncertainty vanishes. Still, our results show that uncertainty has the potential to influence regional growth as it decreases the effectiveness of the EU funds for the host regions. This translates into the fact that the macroeconomic conditions of countries influence the returns of the EU funds received at the local level. Finally, our findings for the control variables are in line with the prior findings. Employment rate, capital per worker, per capita industrial value-added and share of employment with tertiary education positively affect regional per capita GDP growth. On the contrary, lagged per capita income and population density negatively influence regional economic growth. The prior set of findings shows that cohesion policy and regional funds can decrease territorial disparities by spurring growth in relatively less developed regions of the EU. However, our results also point out that rising macroeconomic uncertainty at the country level has local reflections as it decreases the returns of the EU funds at the regional level. In this sub-section, we aim to provide additional evidence that these results are robust to other dimensions. First, we estimate the conditional models by considering two specific cohesion fund categories of the cohesion policy. We replicate the prior analyses for the ERD and ES funds separately. Second, we group the EU regions based on the Objective 1 criteria of the Cohesion Policy and check whether our concern holds across different sets of areas belonging to different development levels. Our first set of robustness results are provided in Tables 4 and 5. We find out that certain EU funds (ERD and ES, respectively) boost regional growth within the EU territory. Once again, non-linear effects show that the positive effect of the ERD and ES funds turns negative after a certain level for the host regions. Additionally, the impact of the per capita GDP and the other regional controls on economic growth aligns with the previous findings. For the two-way models with no interaction term, we observe a stronger effect for the uncertainty index once we consider the ES funds in Table 5. However, once we consider the potential interplay between macroeconomic uncertainty and the EU funds, this effect vanishes for both ERD and ES funds (Tables 4 and 5). Moreover, similar to earlier findings, the interaction term is not significant for the two-way models. Yet, the interaction term is negative and jointly significant with the EU funds and uncertainty. These results validate once again that the positive impact of the ERD and ES funds declines with rising country-wise macroeconomic uncertainty. A second check we carry out is related to the changing nature of EU funds and the interplay between funds and macroeconomic uncertainty once regional heterogeneity is considered. We group the EU regions based on the objective criteria of the EU cohesion Table 5 Impact of the EU funds and macroeconomic uncertainty on regional growth (Robustness analyses for ES Funds). (1) (2) (3) (4) (5) (6) (7) (8) Per capita GDP (t-1) −0.2670 a −0.2417 a −0.2728 a −0.2425 a −0.3426 a −0.2516 a −0.3460 a −0.2527 a (0.0222) (0.0238) (0.0228) (0.0239) (0.0275) (0.0236) (0.0279) (0.0238) PC EU funds 0.0025 0.0041 a 0.0068 a 0.0049 a 0.0027 0.0040 a 0.0062 a 0.0051 a (0.0019) (0.0015) (0.0022) (0.0018) (0.0019) (0.0015) (0.0023) (0.0018) PC EU funds_sqr. −0.0012 a −0.0009 a −0.0013 a −0.0009 a −0.0011 a −0.0009 a −0.0012 a −0.0010 a (0.0004) (0.0003) (0.0004) (0.0003) (0.0004) (0.0003) (0.0004) (0.0003) Uncertainty Index 0.0021 −0.0411 a 0.1531 a −0.0117 −0.0455 a −0.0386 a 0.0769 b −0.0006 (0.0144) (0.0138) (0.0420) (0.0381) (0.0144) (0.0142) (0.0393) (0.0380) Employment rate 0.1387 a 0.1008 a 0.1377 a 0.1009 a 0.2336 a 0.1412 a 0.2312 a 0.1417 a (0.0408) (0.0351) (0.0403) (0.0351) (0.0434) (0.0384) (0.0428) (0.0383) Capital per worker (in ln) 0.0144 a 0.0264 a 0.0128 a 0.0260 a 0.0307 a 0.0264 a 0.0290 a 0.0258 a (0.0046) (0.0042) (0.0045) (0.0042) (0.0049) (0.0043) (0.0048) (0.0043) Industrial value added per capita (in ln) 0.1449 a 0.0948 a 0.1462 a 0.0952 a 0.1438 a 0.0956 a 0.1449 a 0.0962 a (0.0126) (0.0112) (0.0125) (0.0112) (0.0129) (0.0111) (0.0129) (0.0112) Population density (in ln) −0.0140 −0.0767 a −0.0213 −0.0776 a −0.1277 a −0.0933 a −0.1318 a −0.0945 a (0.0363) (0.0269) (0.0362) (0.0267) (0.0377) (0.0279) (0.0374) (0.0277) PC EU fund b Uncertainty Index −0.0545 a −0.0105 −0.0440 a −0.0136 (0.0139) (0.0122) (0.0134) (0.0122) Tertiary Education 0.2675 a 0.1101 a 0.2635 a 0.1110 a (0.0278) (0.0300) (0.0278) (0.0301) Test for joint sign. (interaction and PC EU funds) 8.18 4.01 5.97 3.97 [0.00] [0.02] [0.00] [0.02] Test for joint sign. (interaction and Uncertainty 7.66 5.15 9.85 4.66 [0.00] [0.00] [0.00] [0.01] Observations 3318 3318 3318 3318 3281 3281 3281 3281 R-squared 0.3223 0.5286 0.3273 0.5288 0.3797 0.5314 0.3829 0.5317 Number of regions 196 196 196 196 196 196 196 196 Regional controls Yes Yes Yes Yes Yes Yes Yes Yes Year FE No Yes No Yes No Yes No Yes Cross-section FE Yes Yes Yes Yes Yes Yes Yes Yes Notes: Clustered (NUTS-2) standard errors in parentheses. **p <0.05. a p<0.01. b p<0.1. M. Pinar and B.C. Karahasan The Journal of Economic Asymmetries 30 (2024) e00382 9 policy. Objective 1 group consists of the regions with a per capita GDP below 75% of the average per capita GDP of the EU. We estimate the augmented models for those regions assigned in the Objective 1 group in Table 6 and the non-Objective 1 group in Table 7. We find out that the positive impact of EU funds is interestingly weaker for the Objective 1 regions. This highlights the potential that there is still more room for improving the sub-channels that will expand the impact of the cohesion policy for the less developed EU regions. More centrally, we find out that the interaction term is negative and strongly significant for the two-way models. Note that the interaction terms are negative but not significant for the two-way fixed effect models for the non-Objective 1 regions. In any case, joint significance test results indicate that interaction term and EU funds are jointly significant. Overall these results confirm that macroeconomic uncertainty decreases the effectiveness of the EU funds for EU regions regardless of their development level. 5. Conclusion The effect of the cohesion policy on regional growth has been a long-lasting debate for understanding the success of the EU integration since the early policy implementations. While scholarly literature has conflicting results on the potential contribution of the cohesion policy to regional growth, there is a policy consensus on the efficiency of the cohesion policy to boost territorial integration. Meanwhile, there is growing interest among the science community in exploring the potential moderating effects and conditions for the sound working of the cohesion policy. In this study, we explore an uninvestigated area and examine whether macroeconomic uncertainty at the country level causes any efficiency loss in the applicability of the EU cohesion policy. Overall, we find that rising macroeconomic uncertainty diverts the long-term growth trajectories of the regions and decreases the efficiency of the EU funds received by the host regions. In other words, regions located in countries with higher uncertainty benefit less from the EU funds provided under the cohesion policy framework. In return, these regions fall behind the others in terms of regional growth and development. Our paper has various policy implications. First, country-level uncertainty makes the EU funds less effective. Therefore, policymakers responsible for the allocation of funds could allocate resources across the regions based on the uncertainty levels of the countries to have a more effective allocation of funds to promote economic growth. In other words, some of the funds could be conditional on country-level uncertainty to increase the effectiveness of the funds. Second, countries with more regional disparity problems should not neglect the overall macroeconomic conditions. Otherwise, potential negative externalities will limit the impact of the EU cohesion policy at the local level. Therefore, countries that are candidates to receive higher amounts of EU funds must Table 6 Impact of the EU funds and macroeconomic uncertainty on regional growth (Robustness analyses for Objective 1 Regions). (1) (2) (3) (4) (5) (6) (7) (8) Per capita GDP (t-1) −0.2100 a −0.1720 a −0.2125 a −0.1730 a −0.2515 a −0.1876 a −0.2528 a −0.1889 a (0.0298) (0.0324) (0.0303) (0.0324) (0.0428) (0.0369) (0.0432) (0.0370) PC EU funds 0.0112 b 0.0055 0.0139 a 0.0085 c 0.0122 b 0.0067 0.0136 b 0.0091 c (0.0045) (0.0044) (0.0046) (0.0045) (0.0051) (0.0047) (0.0053) (0.0049) PC EU funds_sqr. −0.0018 a −0.0003 −0.0016 a −0.0003 −0.0022 a −0.0006 −0.0021 a −0.0005 (0.0005) (0.0006) (0.0005) (0.0006) (0.0005) (0.0006) (0.0005) (0.0006) Uncertainty Index 0.0322 −0.0000 0.3893 a 0.2867 a 0.0319 0.0220 0.2359 c 0.2727 b (0.0252) (0.0271) (0.1151) (0.1061) (0.0274) (0.0267) (0.1279) (0.1108) Employment rate 0.2239 a 0.1346 b 0.2230 a 0.1348 b 0.3479 a 0.2056 a 0.3456 a 0.2040 a (0.0675) (0.0612) (0.0671) (0.0604) (0.0818) (0.0738) (0.0818) (0.0734) Capital per worker (in ln) 0.0097 c 0.0105 c 0.0067 0.0078 0.0158 b 0.0087 0.0137 c 0.0063 (0.0049) (0.0056) (0.0049) (0.0057) (0.0076) (0.0076) (0.0078) (0.0077) Industrial value added per capita (in ln) 0.1080 a 0.0738 a 0.1095 a 0.0756 a 0.1187 a 0.0867 a 0.1200 a 0.0885 a (0.0146) (0.0133) (0.0148) (0.0133) (0.0213) (0.0180) (0.0216) (0.0181) Population density (in ln) −0.0648 c −0.0092 −0.0635 c −0.0085 −0.1257 b −0.0620 c −0.1196 b −0.0534 (0.0367) (0.0363) (0.0368) (0.0365) (0.0542) (0.0371) (0.0547) (0.0363) PC EU fund c Uncertainty Index −0.0713 a −0.0577 a −0.0404 −0.0502 b (0.0227) (0.0204) (0.0254) (0.0214) Tertiary Education 0.1880 a 0.1554 b 0.1845 a 0.1503 b (0.0468) (0.0694) (0.0469) (0.0691) Test for joint sign. (interaction and PC EU funds) 7.6 4.75 3.56 3.45 [0.00] [0.01] [0.03] [0.04] Test for joint sign. (interaction and Uncertainty 6.03 4.03 2.2 3.05 [0.00] [0.02] [0.12] [0.05] Observations 1467 1467 1467 1467 1185 1185 1185 1185 R-squared 0.2523 0.4350 0.2577 0.4383 0.3275 0.4658 0.3295 0.4687 Number of regions 86 86 86 86 77 77 77 77 Regional controls Yes Yes Yes Yes Yes Yes Yes Yes Year FE No Yes No Yes No Yes No Yes Cross-section FE Yes Yes Yes Yes Yes Yes Yes Yes Notes: Clustered (NUTS-2) standard errors in parentheses. a p<0.01. b p<0.05. c p<0.1. M. Pinar and B.C. Karahasan