Analysis of spatial dependencies and spatial effects in the relationship between economic growth and unemployment in Europe 163DOI: 10.15201/hungeobull.74.2.3 Hungarian Geographical Bulletin 74 (2025) (2) 163–176. Introduction Unemployment and insufficient economic growth are among the most important eco- nomic problems for every country nowa- days. Policymakers and society concurrently monitor both processes as the primary indi- cators of economic development. Increasing the number of unemployed people concerns both developing and developed economies. However, unemployment in developing countries is primarily the result of capital inadequacy. In turn, technological progress is the reason for the decrease in the num- ber of employed people in developed states (Soylu, Ö.B. et al. 2018). Another difference between unemployment in developed and developing countries is that the former ex- perience extended periods of rapid economic growth. In contrast, the poorer ones either have never experienced significant growth or have had periods of economic growth and decline that occurred alternately (Madito, O. and Khumalo, J. 2014). In theoretical and empirical macroeconom- ic research, the linkage between economic growth and the unemployment rate is widely studied. This relationship is one of the most important in economics. It is widely accept- ed that a higher growth rate of the Gross 1 Nicolaus Copernicus University in Toruń, Faculty of Economic Sciences and Management, The Department of Applied Informatics and Mathematics in Economics. Gagarina 13a, 87-100 Toruń, Poland. E-mail: m.jankiewicz@umk.pl Analysis of spatial dependencies and spatial effects in the relationship between economic growth and unemployment in Europe Mateusz JANKIEWICZ 1 Abstract The study presents an analysis of the relationship between unemployment and economic growth in European countries. The significant influence of the labour market situation on the economic condition of states is well- known. The analysis in this research was conducted using data from 43 selected European countries from 2006 to 2019. To evaluate the relationship between economic growth and the unemployment rate, enriched with spatial dependencies, spatial models for pooled time series and cross-sectional data (TSCS) were estimated. The neighbourhood was quantified using three types of connection matrices: (1) based on the common border criterion, (2) based on the maximum distance criterion, and (3) based on the similarity in economic situations. Matrices (1) and (2) relate to the geographical neighbourhood, while matrix (3) defines the economic neighbour- hood based on the values of the consumption expenditures per capita. The choice of these types of matrices was associated with the migration process (geographical neighbourhood) and the imitation effect of labour market strategies (economic neighbourhood) mentioned above. Based on the estimation and verification results of the Spatial Durbin Models (SDM), the spatial spillover effects were evaluated. Cumulative spatial effects allowed us to determine countries with the greatest influence on others and countries that are following the leading ones. Keywords: economic growth, economic neighbourhood, Spatial Durbin Model, short-term spatial effects, unemployment Received October 2024, accepted May 2025. Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176.164 Domestic Product (GDP) of an economy caus- es a decrease in the unemployment rate. The negative short-run dependence between these processes was named Okun’s law, which was formulated by Arthur Melvin Okun (Okun, A.M. 1962). Consequently, Okun pointed out that labour market modifications are the re- sults of changes in economic development. The dependent variable in the most common direct regression of Okun’s law is the differ- ence between the actual unemployment rate (marked uit) and the natural unemployment rate (or the equilibrium rate of unemployment – marked as uit*). This process is dependent on the difference between actual (yit) and poten- tial or equilibrium output (yit*). The regres- sion parameter between these two processes is marked as β. Index i refers to the region (or another territorial unit), while index t indi- cates time. Moreover, β is an Okun’s coeffi- cient (ultimately with a negative sign). This method of considering Okun’s relationship is called the Trial Gaps. Previous studies show that Okun’s law is stable in many countries (Ball, L. et al. 2017). Instability in the relation- ship between states is visible in terms of the economic slowdown, resulting, i.e. from the financial crisis (Cazes, S. et al. 2011). Apart from the Trial Gaps method, there are two alternative methods to estimate Okun’s coefficient: (i) the First Differences method, and (ii) the Fitted Trend and Elasticity method (Barreto, H. and Howland, F. 1993). In the First Differences method, the form of the model is similar to that in the Trial Gaps method. However, the natural unemployment rate and potential output are replaced by the time lags of the actual unemployment rate (uit-1) and actual output (yit-1), respectively. As a result, the first differences of the unemployment rate and Gross Domestic Product are the dependent and independent variables, respectively. In turn, in the Fitted Trend and Elasticity meth- od, the relationship between the natural loga- rithms of the unemployment rate (dependent variable) and GDP (independent variable) is considered. Moreover, the model is comple- mented by the deterministic time trend. Based on the shown relationship, it is pos- sible to answer what level of GDP should be expected under the economic conditions during the sample period if a certain level of unemployment is given. However, the reversed dependence has to be considered (Barreto, H. and Howland, F. 1993). The inverse relationship in the literature is also called Okun’s law. The second approach is closer to economic growth theory, where the labour force is one of the factors that influ- ence the production level. In this research, the economic growth changes as a result of changes in unemployment are speculated. In general, Okun’s law has been formulat- ed for closed economies and refers to region- al analyses. Nevertheless, nowadays, when we are dealing with the free movement of the labour force and capital, the law men- tioned above can also be used in analyses across countries. As concluded by Palombi, S. et al. (2015) “(…) the reliability of Okun’s coefficients is not only of paramount impor- tance for macroeconomic policy, but also for the regional distribution of unemployment rates in an open spatial system”. Hence, we are able to recognize that the set of territori- al units which are open economies creates an open spatial system. In this case, the spatial dependencies between countries in the study of the relationship between un- employment rates and economic growth are also significant. Formánek, T. and Hušek, R. (2016) pointed out that the spatial con- nections shown during the analysis of the link between labour market conditions and economic growth can also be interpreted as the influence of real, practically unobserv- able, and difficult-to-quantify spatial effects. Among these effects, they mention cross-bor- der work commuting preferences, account- ing for administrative employment barriers between countries, language differences, and aerial distances vs. topology. The importance of including the spatial connections in the analysis of the topic mentioned above across countries can be explained with the push-pull concept of territorial mobility as well (Lee, E.S. 1966). Based on this idea, every coun- 165DOI: 10.15201/hungeobull.74.2.3 Hungarian Geographical Bulletin 74 (2025) (2) 163–176. try has characteristics that are conducive to taking up employment (for example, higher earnings, higher levels of work culture) as well as determinants that are unfavourable. Hence, labour force migration causes modifi- cations in the national labour market, and at the same time, in the labour market abroad, as well as in the output of other states. Moreover, when the labour force of a country increases (for example, in times of positive economic changes), there can be a problem with providing jobs for all, and as a result, people decide to look for jobs abroad (Maza, A. 2022). Based on all these determinants, the spatial aspect in the analysis of the relation- ship between unemployment and economic growth is desirable. This paper provides in- formation about the spatial spillovers deter- mined across countries, which constitutes the novelty and research gap. In this study, the relationship between the unemployment rate and economic growth in 43 selected European economies between 2006 and 2019 is considered. The aim of the research is to designate the consequences for countries’ economic growth caused by the modifications in the labour market through changes in the unemployment rate, deter- mining spatial spillover effects. Cumulative spatial effects allowed us to identify coun- tries with the greatest influence on others and countries that are following the leading ones. This analysis allows us to formulate some recommendations for governments in order to use the cooperation between neigh- bouring countries to improve their own la- bour market conditions and, as a result, the economic development. Almost all studies analysing economic growth and unemploy- ment in Europe deal with the European Union members only. In this research, it was important to include countries located in Eastern Europe (considered relatively less developed than the EU members) as well. This allows for a wider analysis of the sub- ject of Okun’s relationship in Europe. In turn, the period from 2006 to 2019 is the longest available period with complete data for the chosen countries. In the formation of the crucial economic indicators, such as economic growth and unemployment, long-term tendencies con- taining the crisis periods are important. In the chosen period, the financial crisis from 2008 and economic slowdown from 2012 are included. Given that countries, especially neighbouring ones, cooperate nowadays, the spatial, and spatio-temporal dependencies in the presented investigation are includ- ed. This allows for evaluating the impact of specific changes in a given region on the economic situation in others. Therefore, the spatio-temporal Durbin Model is estimated, based on which the spatial spillovers are de- termined. The hypotheses of this study are as follows: (1) changes in the labour market in given country have a significant impact on the output of the other, (2) changes in the labour market conditions in more developed countries have a stronger influence on the economic growth in others than in poorer ones, (3) the economic similarity of territo- rial units is more important than their geo- graphical proximity in the formation of the relationship between economic growth and unemployment. Review of the scientific literature In the literature, the relationship between the unemployment rate and economic growth is widely speculated. Almost all researchers consider this dependence based on Okun’s law. In previous studies, it is tough to find analyses using Okun’s law in the Fitted Trend and Elasticity approach. The first type of study conducted on the relationship between these processes con- tains analyses for a single country based on time series. Valadkhani, A. and Smyth, R. (2015) considered this relationship for the US in the period of 1948–2015 using quarterly data. They estimated models for the Trial Gaps method of Okun’s law, determining potential output and the natural rate using the Hodrick-Prescott filter (Hodrick, R.J. and Prescott, E.C. 1997). Binet, M.E. and Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176. Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176.166 Facchini, F. (2013) speculated models in the Trial Gap version for French regions in the years 1990–2008. They concluded about re- gional heterogeneity based on Okun’s coeffi- cient estimation. In turn, Li, C.S. and Liu, Z.J. (2012) used a Vector Error Correction Model (VECM) and Granger causality test to inves- tigate the long-term and short-term depend- encies between unemployment and econom- ic growth in China from 1978 to 2010, among others. The Granger causality approach is a very popular tool in the analyses concern- ing these processes. For example, Louail, B. and Riache, S. (2019) used this method to speculate about unemployment and eco- nomic growth in Saudi Arabia in the period 1991–2017. In turn, Alhdiy, F.M. with co-au- thors pointed out that changes in unemploy- ment caused changes in the GDP per capita for Egypt in the years 2006–2013 (quarterly data used) (Alhdiy, F.M. et al. 2015). Besides, Sadiku, M. with co-authors, studied causality using the Granger approach between these processes in North Macedonia (Sadiku, M. et al. 2015). The case of the verification of Okun’s law for Nigeria, South Africa, and the USA was of interest to Onakoya, A.B. and Seyingbo, A.V. (2020). Based on the First Difference approach, they concluded that for Nigeria, the law formulated by Arthur Okun is not applicable. Moreover, this major macroeconomic relationship was verified for developing countries like Jordan (Al-Habees, M.A. and Rumman, M.A. 2012), Albania (Nikolli, E. 2014), and South Africa (Madito, O. and Khumalo, J. 2014; Makaringe, S.C. and Khobai, H. 2018). All authors cited so far have not consid- ered the spatial dependence between terri- torial units, which is a crucial issue in terms of international cooperation. Villaverde, J. and Maza, A. (2016) used the spatial panel approach to examine Okun’s law for Spanish regions from 2000 to 2014. They employed various types of connection matrices, pri- marily distance matrices, citing Tobler’s First Law of Geography (Tobler, W.R. 1970). Based on the Spatial Durbin Model, Villaverde, J. and Maza, A. (2016) quanti- fied the direct and indirect effects of output growth. Montero Kuscevic, C.M. (2014) also quantified spatial spillovers in the relation- ship between output and unemployment rate at the metropolitan statistical area level (MSA) in the United States. Pereira, R.M. (2014) focused on the same statistical area to estimate regional spillovers, with a particular emphasis on the asymmetry in Okun’s law. Additionally, Oberst, C. and Oelgemöller, J. (2013), as well as Salvati, L. (2015), high- lighted the spatial dependencies and re- gional effects in the relationship between unemployment and output growth. The first study examined the mentioned dependence for German regions, while the second study used data from Italian provinces. Basistha, A. and Kuscevic, C.M.M. (2017), Palombi, S. et al. (2017), Elhorst, J.P. and Emili, S. (2022), and Jankiewicz, M. (2023) emphasized the important role of spatial connections in ana- lyzing the relationship and also quantified spatial spillovers based on estimated models of Okun’s relationship. Nevertheless, they determined only cumulative direct and in- direct spatial effects additionally at the re- gional level. This analysis shows the strength of the spatial spillovers for individual ter- ritorial units at the country level. Detailing the spatial spillover effects and the transfer of the analysis to the macroeconomic scale are the novelties introduced in this study. Okoro-Ugochukwu, N.A. and Adenomon, M.O. (2021) conducted their research on the relationship between unemployment and economic growth in Nigerian regions, also utilizing a spatial approach. In turn, Xu, B. et al. (2021) introduced spatial dependence in the analysis of the relationship between entrepreneurship and regional economic growth across China provinces from 2010 to 2016. In their research, they used Mixed Geographically Weighted Panel Regression with Spatial Autoregression (MGWPR-SAR). In all cited works, the spatial dependence in the analysis of Okun’s relationship turned out to be statistically significant. This study employs different types of spatial connection matrices, which can provide further insights 167DOI: 10.15201/hungeobull.74.2.3 Hungarian Geographical Bulletin 74 (2025) (2) 163–176. into the verification of Okun’s law. Two ma- trices are based on geographical proximity: the first uses the common border criterion, while the second is built based on the maxi- mum distance criterion (two countries are neighbours when the distance between them does not exceed 1000 kilometres). The geo- graphical proximity is the most often consid- ered in spatial analyses. This is related to the First Law of Geography formulated by Tobler, W.R. (1970). The third proximity matrix is based on economic simi- larity using the consumption expenditures per capita values. The premise for using this type of neighbourhood matrix is the sig- nificant role pointed out in the literature of consumption in the formation of economic growth and unemployment. Consumption is considered the main factor driving the economy. Moreover, the increase in con- sumption expenditures leads to a reduction in the unemployment rate. This is the very first time the proximity matrix based on the consumption expenditures level is employed in research based on Okun’s law, which con- stitutes the next added value in the scientific literature. Research methodology The first part of the investigation deals with the analysis of a spatio-temporal structure of economic growth and unemployment to detect long-term tendencies in their formation. Initial- ly, the spatio-temporal trend, which is respon- sible for the non-stationarity in the average, is analysed. The general form of the spatio-tem- poral trend model is characterized as follows: where Z denotes a considered process, si = [xi, yi] is a vector of spatial units coordinates (i = 1,2,…, N denotes the number of territorial unity), k + m + l ≤ p, and t indicates time. Next, the spatial autocorrelation as the second component of the spatial struc- ture, characterizing dependencies between neighbouring units, is tested. The Moran’s I statistic verifying the presence of global spatial autocorrelation is described with the following formula (Moran, P.A.P. 1948; Schabenberger, O. and Gotway, C.A. 2005): where yi indicates the observation of the process in the ith region, ӯ denotes the average value of the process, and W is the spatial connections ma- trix between units. In this study, three types of spatial connections matrix are concerned. The first of them is the most used in spatial analyses a neighbourhood matrix based on the common border criterion (marked with W1). The other two matrices are created using the maximum distance criterion, one of which concerns a geo- graphic distance (signed as W2), but the next is built based on the economic distance (named as W3). According to the W2 matrix, two regions are neighbours when the distance between them does not exceed 1000 kilometers. In turn, the economic distance in the W3 matrix is calculat- ed using the level of consumption expenditures per capita. The W3 matrix is constructed using the algorithm presented in the article conducted by Jankiewicz, M. and Szulc, E. (2021). To de- termine the economic spatial weights matrix, the Euclidean distance between pairs of countries is calculated (in case of only one variable, this is the Manhattan distance). Then, the borderline value of the distance between countries is fixed, and the values exceeding the borderline are re- placed with zeros. Subsequently, the non-zero elements are transformed by inversion. Finally, the elements of the proximity matrix are row standardized to one. In the next step of the study, for processes filtered out from the long-term tendencies, the spatio-temporal model of dependence is estimat- ed. The general form of the model is as follows: (1) (2) (3) Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176. Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176.168 where Yi,t denotes the GDP per capita, where- as X1i,t and X2i,t are the unemployment rate and inflation, respectively, in ith country in time t. In turn, α, β1 and β2 are the structural parameters of the model, and εi,t indicates the spatio-temporal random component. To confirm the need to include the spatial dependencies in the model (3), the Lagrange Multiplier tests (LM) in two versions – basic and robust – are used (Anselin, L. et al. 2004). Consequently, the spatio-temporal Durbin model is considered in the following form (Elhorst, J.P. 2011): where wij,t are elements of the block matrix of spatio-temporal connections which is ex- pressed as (Szulc, E. and Jankiewicz, M. 2018): wherein W1, W2,…,WT are standard spatial connectivity matrices, such as in (2). These matrices are the same for all years in the case of W1 and W2. In turn, the economic distance matrix (W3) changes over time. Based on the model (4), the so-called direct and indirect spatial effects can be quantified. To obtain the mentioned effects, the gener- al form of the non-dynamic model, i.e., the model (6): should be transformed to its reduced form (Vega, S.H. and Elhorst, J.P. 2013): Excluding from the matrix Xt the vector corresponding to the variable Xk, i.e., Xkt the following form is obtained: where Xt indicates the matrix from which the Xkt has been removed. The matrix of partial derivatives of Y con- cerning the kth explanatory variable of X in region 1 up to region N at a particular point in time, given as (9), designates the short- term effects. The diagonal elements of matrix (9) rep- resents direct effects. In turn, the non-diago- nal elements correspond to spillover effects. Direct effects define impacts of change in ob- servation xk for ith spatial unit (marked with xik) on the values of the depending variable in the same region (yi). With non-diagonal elements of effects matrices, two types of impacts (spatial spillovers) can be identified (Lesage, J. and Pace, R.K. 2009): (1) Average Impact to an Observation – the average impact on individual observation yi resulting from changing kth explanatory variable by an amount across all observations (the average of the sum across the ith row), (2) Average Impact from an Observation – the average impact over all yi from changing the kth explanatory variable by an amount in the jth observation (the average of the sum down the jth column). Data and empirical results This study examines the relationship be- tween economic development and unem- ployment in general. Economic growth is measured by Gross Domestic Product (GDP) per capita (variable Y). Unemployment is quantified as the percentage of unemployed individuals with- in the total labour force (variable X1). (4) (6) (7) (8) (9) (5) . 169DOI: 10.15201/hungeobull.74.2.3 Hungarian Geographical Bulletin 74 (2025) (2) 163–176. Based on the literature review, the analysis also includes the inflation rate (variable X2) as an additional factor influencing the eco- nomic situation of each country. In partic- ular, the study investigates the spatial, and spatio-temporal relations between the anal- ysed processes. All data are obtained from the World Bank database (https://data.world- bank.org/indicator) and directly used. The following indicators were used: GDP per capita (current USD) – variable X1, unem- ployment (% of total labour force) – variable , and inflation measured by the consumer price index (the annual percentage change in the cost to the average consumer of acquiring a basket of goods and services that may be fixed or changed at specified intervals) – vari- able X2. Data refer to the period of 2006–2019. All calculations and figures are performed using R software (version 4.1.1). Table 1 presents the descriptive statistics of two crucial variables in this analysis – Gross Domestic Product (Y) and unemployment (X1) over the period 2006–2019. It is worth noting that in the formation of both phenomena, the median value is lower than the mean value. In combination with the skewness coefficient values (1.3235 and 1.5696 for GDP and unem- ployment, respectively), we can note that the distributions of the considered variables are skewed right. This means that more than half of the observations have values lower than the mean. The volatility of the variables is simi- lar and moderate. This is evidenced by the values of volatility index around 60 percent. Moreover, the kurtosis coefficient, with a val- ue higher than two for both phenomena, indi- cates a too peaked distribution in comparison with the normal distribution. Additionally, in Figure 1, the scatterplot of the dependence be- tween unemployment and economic growth for all countries in the whole research period (N = 602) is presented. The included regres- sion line of negative slope allows us to pre- Table 1. The descriptive statistics of variables used in the estimated models Variable Mean Median Minimum Maximum Y X1 34,263.0000 9.8093 30,409.0000 7.6550 7599.5000 2.0100 114,890.0000 36.0300 Variable Standard deviation Volatility index, % Skewness Kurtosis Y X1 20,028.0000 6.2291 58.46 63.50 1.3235 1.5696 2.7865 2.2923 Source: Compiled by the author. Fig. 1. The scatterplot of dependence between unemployment rate and economic growth in Europe in the period of 2006–2019 (N = 602). Source: Authors’ own elaboration. Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176. Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176.170 sume that in the considered area, higher val- ues of unemployment are linked with lower economic growth. In the first step of the research, the spa- tio-temporal trend models for GDP per capita and the unemployment rate were considered. The results of estimation and verification of the models mentioned above are presented in Table 2. The estimates of parameters θ100 and θ010 (negative and positive, respectively) show that in the examined period, the GDP per capita values have been increasing in the north-western direction. In turn, the average higher unemployment rate in the period was observed in South-eastern Europe. Moreover, the significant parameter θ001 for the GDP and also the p-values of the parameters θ001 and θ002 for the unemployment confirm the rel- evant changes of these processes over time. For both models, residual spatial autocor- relation occurs in light of all defined spatial connection matrices. Nonetheless, the depen- dence between countries is stronger for the variable Y than X1 for . Among the consid- ered matrices, the highest value of Moran’s I is obtained for geographical neighbours with a common border (W1), indicating the strongest reliance. But the strength of depen- dence between countries with similar values of consumption expenditures (W3) for GDP per capita does not differ significantly from that obtained for common border proximi- ty. The geographical distance matrix shows the weakest connections between regions in terms of both GDP per capita and unemploy- ment rate. On the other hand, the economic distance is not more significant for the un- employment rate than the physical distance. Spatial dependence analysis began with the determination of its nature. For this purpose, a classical spatio-temporal mod- el of the relationship between variables Y and X1 was estimated and verified (Table 3). Additionally, the influence of inflation is con- sidered (variable X2). Estimates of significant parameters β1 and β2 indicate that increased unemployment and inflation rates harm eco- nomic growth. The negative sign of Okun’s parameter (β1) confirms the truth of the ma- jor macroeconomic relationship in European countries. Moreover, the significant Moran’s statistics indicate the presence of spatial au- tocorrelation in the model residuals. Table 4 shows the results of the estimation and verification of Spatial Durbin Models for the considered relationship. The sign of the estimate of parameter β1 did not change com- pared to the OLS model (β1 for OLS model is equal to -0.3852). Nevertheless, the strength of the dependence is less (the greatest difference is observed using the W1 matrix). Based on the significance of parameter p, it can be conclud- ed that there is a similarity in GDP per capita between neighbouring countries. The most similar regions are those that have a common border, while the least similar are states with a similar level of consumption expenditures per capita. On the other hand, the statistical signif- icance of parameter θ1 indicates that changes in the labour market in neighbouring countries Table 2. The results of spatio-temporal trend models estimation and verification Parameter Y X1 Estimate (p-value) Estimate (p-value) θ000 θ100 θ010 θ001 θ002 9.2835 (0.0000) -0.0250 (0.0000) 0.0269 (0.0000) 0.0162 (0.0001) – 3.3423 (0.0000) 0.0037 (0.0086) -0.0325 (0.0000) 0.1302 (0.0000) -0.0094 (0.0000) Moran test W1 W2 W3 W1 W2 W3 I p-value 0.3200 0.0000 0.1821 0.0000 0.3195 0.0000 0.2283 0.0000 0.0801 0.0007 0.1119 0.0002 Source: Compiled by the author. 171DOI: 10.15201/hungeobull.74.2.3 Hungarian Geographical Bulletin 74 (2025) (2) 163–176. (regardless of the used spatial connections matrix) significantly influence the output in a certain state. The impact has the same nature as changes in the unemployment rate in a cer- tain country. In case of geographical proximity (matrices W1 and W2), it is the impact of weak- er strength. In contrast, in the economic neigh- bourhood (matrix W3), the influence of labour market changes in other countries is stronger than changes within the given country. The statistical significance of parameters p (a close significance for W3 matrix) and θ1 allows for quantifying short-term indirect spatial effects caused by shocks in the labour market condi- tions in European countries. The desirable property of all models is the absence of spatial autocorrelation in residuals. Based on the Akaike criterion (AIC) and the logarithm of likelihood (Log-lik) values, the Spatial Durbin Model with the neighbourhood quantified using the W3 matrix is the best. Figure 2 presents the spatial distributions of spatial spillovers obtained when considering a common border neighbourhood. The top map in Figure 2 shows the distribution of average inflows via the unemployment rate in individ- ual countries on the output growth in a given country. We can see that most of the Western and Northern European regions (Finland, Ireland, Norway, Sweden, United Kingdom) were among those that received transmission impulses from other regions with the lowest strength. It should be noted that these are countries characterized by rel- atively high economic growth. The countries most affected by all other countries through the transmission of labour market conditions included the Czech Republic, Estonia, Denmark, Hungary, Lithuania, Moldova, and Switzerland. Map b) of Figure 2 contains the distribution of the average impacts of a given country’s unemployment rate on the eco- nomic growth in all other econ- omies. It is worth noting that most of the countries least in- fluenced by others were those that strongly affected other countries. Thus, changes in the unemployment rate in France, Germany, Serbia, Turkey, and Russia most strongly affect the output in other countries. On the other hand, countries such as the Czech Republic, Denmark, Estonia, Italy, Moldova, Portugal, Spain and Sweden gave the weakest transmission impulse to other countries. In turn, Figure 3 shows the distributions of indirect effects evaluated based on the geographical distance dependence between countries. As with Figure 2, the top map Table 3. The results of the OLS Okun’s relationship model estimation and verification Parameter Estimate t-statistics p-value α β1 β2 0.0632 -0.3852 -0.0185 3.7480 -13.0850 -6.6830 0.0002 0.0000 0.0000 Diagnostics R2 0.2484 Moran test W1 W2 W3 0.1801 (0.0000) 0.0795 (0.0007) 0.0881 (0.0023) LMerr 33.8150 (0.0000) 10.6239 (0.0011) 7.6657 (0.0056) LMlag 68.0270 (0.0000) 24.9503 (0.0000) 43.0832 (0.0000) RLMerr 21.2290 (0.0000) 9.7046 (0.0018) 77.4523 (0.0000) RLMlag 55.4420 (0.0000) 24.0311 (0.0000) 112.8698 (0.0000) Source: Compiled by the author. Table 4. The results of estimation and verification of the non-dynamic Spatial Durbin models Parameter Model SDM W1 SDM W2 SDM W3 α β1 θ1 β2 ρ 0.0430 (0.0049) -0.2800 (0.0000) -0.1724 (0.0010) -0.0148 (0.0000) 0.3854 (0.0000) 0.0654 (0.0001) -0.3514 (0.0000) -0.1383 (0.0381) -0.0162 (0.0000) 0.2224 (0.0005) 0.0568 (0.0003) -0.3112 (0.0000) -0.4835 (0.0000) -0.0148 (0.0000) 0.0818 (0.0541) Diagnostics Moran test Log-lik AIC 0.0045 (0.4231) -162.8580 337.7200 0.0004 (0.4655) -194.6190 401.2400 -0.0387 (0.1215) -160.4491 332.9000 Source: Compiled by the author. Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176. Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176.172 contains average impacts to an observa- tion, while the bottom map contains aver- age impacts from an observation. Countries in Central Europe received the strongest transmission impulses from other countries. Moreover, Eastern and Northern European countries were highly sensitive to changes in the unemployment rate in other countries. Simultaneously, most of them had a poor ef- fect on other economies. Among the select- ed countries, some showed weak strength in both spatial impacts, such as Ireland, Norway, and Spain. It is worth noting that most of the Central-Southern European countries (especially Western Balkans econ- omies) were relatively most influenced by others. This group of regions also included Belgium, Latvia, and Lithuania. Figure 4 presents the distributions of short- term indirect effects resulting from the eco- nomic distance between countries. We can see that, in general, countries in Eastern Europe received the highest strength trans- mission impulses from other countries, par- Fig. 3. For explanation see Fig. 2, but the values of (A) and (B) are based on the W2 matrix. Source: Authors’ own elaboration. Fig. 2. The distribution of the average short-term impacts across Europe in the period of 2006–2019 of the spatially lagged unemployment rate on the output growth in individual economies (A), and a change of the unemployment rate in a particular economy on the economic growth in all other economies (B), based on the W1 matrix. Source: Authors’ own elaboration. 173DOI: 10.15201/hungeobull.74.2.3 Hungarian Geographical Bulletin 74 (2025) (2) 163–176. ticularly Belarus, Estonia, Latvia, Lithuania, Moldova and Ukraine. Changes in the un- employment rate in other economies had the most influence on output growth in sever- al Southern European countries, including Croatia, Georgia Greece and Portugal. It is worth noting that the regions that were least influenced by others were the ones that most strongly affected other regions. Conclusions The relationship between the unemployment rate and economic growth, as one of the ma- jor relationships in economics, is widely con- sidered in many studies. Almost all confirm the nature of this relationship, formulated by Arthur Okun, which points out that an increase in the unemployment rate causes a decrease in output growth. It is no different in this research that examines the case of Eu- ropean economies. Almost all cited studies concluded that the relationship between the unemployment rate and economic growth is negative. Some of them underlined the dif- ferences in the operation of Okun’s law be- tween emerging and developed economies. This is the clue for improving this research in the future by dividing the set of countries into two separate groups. The analysis of spatial dependencies al- lows us to conclude the need to include in- teractions between neighbouring countries in Okun’s relationship models. The use of several spatial connection matrices shows that the geographical neighbourhood (de- fined by a common border or distance be- tween countries) was stronger than the economic neighbourhood adopted in this research (defined as similarity in the level of consumption expenditures). This means that, the third hypothesis of the research has not been confirmed. Therefore, there is a need to look for another economic similarity that may be more relevant in the relationship be- tween the unemployment rate and economic growth than geographical proximity. This re- search shows that the geographical location is very important in the case of formation the economic phenomena. Undoubtedly, the geographical proximity is the source of the imitation effect, where economies follow their neighbours, primarily in terms of con- sumption, but not only. Most of the migrant workforce chooses the nearest countries as the goal of their migration. Additionally, the geographical neighbourhood allows econ- omies to cooperate and establish unions more easily, what leads to their faster de- velopment. Which is why the geographical neighbourhood can be the driver for most Fig. 4. For explanation see Fig. 2, but the values of (A) and (B) are based on the W3 matrix. Source: Authors’ own elaboration. Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176. Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176.174 of economic factors. Nevertheless, changes in economic growth and the unemployment rate in neighbouring regions significantly influence output growth in a certain coun- try, regardless of the neighbourhood matrix adopted. This means that different factors are leading the labour force to take up jobs abroad. Some people prefer to be near their family home, while for others, higher earn- ings and much better working conditions drive migration much further. Undoubtedly, competition in the labour market and trust in the government of the host country are also factors that significantly influence the choice of workplace. Quantified short-term spatial indirect ef- fects show a difference in their distributions depending on the neighbourhood matrix used. Firstly, the knowledge spillover effect leads to a change in the country’s labour market. Producers are able to rapidly trans- form their production systems by learning new technologies from manufacturers in other countries (particularly those similar in socio-economic conditions). As a result, economic growth accelerates, and short-term spatial effects become stronger. Moreover, the mobility of production factors strength- ens the spatial spillovers in the short term as well, mainly labour force mobility. The worsening of the labour market in a given country often forces the population to mi- grate for work to other states. Comparing the spatial spillovers analysis with the studies conducted so far, this analysis confirms the significance of the spatial connections in the considered relationship, as concluded also by Palombi, S. et al. (2017), and Elhorst, J.P. and Emili, S. (2022). Based on the analysis pre- sented in this paper, there is the opportunity to formulate some recommendations for pol- icymakers. First of all, they should provide decent conditions of employment and reduce the negative modifications of the labour mar- ket conditions that have unfavourable conse- quences for output. So the pull factors have to be eliminated. Only then will the labour force not be transferred abroad. Moreover, policy- makers should improve the effectiveness of economic policies and the overall condition of the economy, for example through the de- velopment of technologies and infrastructure in the labour market. All these decisions have relevant geographical consequences, par- ticularly in the socio-economic dimension. The progressing modifications in the labour market and the economic structure can lead to changes in the age structure of the popu- lation and the structure of natural resource use. This study shows that the cooperation between the nearest economies significantly influences the economic situation. In the light of these results, policymakers should strive to the intensification of exchange of views and experiences with the governments of their neighbours. In order to slow down the outflow of human capital and, as a result, in- crease a potential of the national economy, policymakers should also support entities in case of the outplacement process. Mostly in the short term, countries that received transmission impulses from other countries with the least (higher) strength were those more (less) affecting other coun- tries. Moreover, the most highly developed countries in the short term were least sensi- tive to labour market shocks in the neigh- bouring states. Nevertheless, there is not an unambiguous division that points out that relatively highly developed countries have a greater influence on others than the poorer ones. Hence, the second hypothesis provided is not completely true. In further research, it is worth conducting an analysis of the division into relatively poor and relatively rich economies. The di- vision of countries considered in this study according to Human Development Index (HDI) level into developing and developed economies can provide significant conclu- sions about differences in the formation of economic growth resulting from changes in the labour market. Moreover, the research can be enriched with other types of spatial connection matrices. It is also interesting to compare the results obtained in this study with the analysis based on another method of estimation of Okun’s relationship (Trial 175DOI: 10.15201/hungeobull.74.2.3 Hungarian Geographical Bulletin 74 (2025) (2) 163–176. Gap and First Differences methods). It is also worth studying the impact of the COVID-19 pandemic on Okun’s relationship. There was a significant shock for the economies of the world, and relevant changes happened. For sure, in the studied relationship, structural change occurred. But the particular conse- quences of the COVID-19 pandemic can be an element of this study’s improvement. REFERENCES Al-Habees, M.A. and Rumman, M.A. 2012. The re- lationship between unemployment and economic growth in Jordan and some Arab countries. World Applied Sciences Journal 18. (5): 673–680. Alhdiy, F.M., Johari, F., Daud, S.N.M. and Rahman, A.A. 2015. Short and long term relationship between economic growth and unemployment in Egypt: An empirical analysis. Mediterranean Journal of Social Sciences 6. (4): 454–454. https://doi.org/10.5901/ mjss.2015.v6n4s3p454 Anselin, L., Florax, R.J.G.M. and Rey, S. 2004. Advances in Spatial Econometrics. Methodology, Tools and Applications. New York, Springer, https://doi. org/10.1007/978-3-662-05617-2 Ball, L., Leigh, D. and Loungani, P. 2017. Okun’s law: Fit at 50? Journal of Money, Credit and Banking 49. (7): 1413–1441. https://doi.org/10.1111/jmcb.12420 Barreto, H. and Howland, F. 1993. There are Two Okun’s law Relationships between Output and Unemployment. Crawfordsville, USA, Mimeo, Wabash College. Basistha, A. and Kuscevic, C.M.M. 2017. The role of spatial GDP spillovers in state-level Okun’s law. Letters in Spatial and Resource Sciences 10. (3): 353–360. https://doi.org/10.1007/s12076-017-0189-7 Binet, M.E. and Facchini, F. 2013. Okun’s law in the French regions: A cross-regional comparison. Economics Bulletin 33. (1): 420–433. Cazes, S., Verick, S. and Al-Hussami, F. 2011. Diverging trends in unemployment in the United States and Europe: Evidence from Okun’s law and the global financial crisis. Employment Working Papers 106. Geneva, UN ILO. Elhorst, J.P. 2011. Spatial panel data models. In Handbook of Applied Spatial Analysis. Eds.: Fischer, M.M. and Getis, A., Cham, Springer, 377–407. https://doi.org/10.1007/978-3-642-03647-7_19 Elhorst, J.P. and Emili, S. 2022. A spatial economet- ric multivariate model of Okun’s law. Regional Science and Urban Economics 93. 103756. https://doi. org/10.1016/j.regsciurbeco.2021.103756 Formánek, T. and Hušek, R. 2016. Spatial aspects of unemployment in the Visegrad-Group economies. Creative and Knowledge Society 6. (2): 1–12. https:// doi.org/10.1515/cks-2016-0007 Hodrick, R.J. and Prescott, E.C. 1997. Post-war US business cycles: An empirical investigation. Journal of Money, Credit, and Banking 29. (1): 1–16. https:// doi.org/10.2307/2953682 Jankiewicz, M. and Szulc, E. 2021. Analysis of spatial effects in the relationship between CO2 emissions and renewable energy consumption in the context of economic growth. Energies 14. (18): 5829. https:// doi.org/10.3390/en14185829 Jankiewicz, M. 2023. Regional economic growth and unemployment in the European Union: A spatio- temporal analysis at the NUTS-2 level (2013–2019). Hungarian Geographical Bulletin 72. (2): 179–192. https://doi.org/10.15201/hungeobull.72.2.6 Lee, E.S. 1966. A theory of migration. Demography 3. 47–57. https://doi.org/10.2307/2060063 Lesage, J. and Pace, R.K. 2009. Introduction to Spatial Econometrics. New York, Chapman and Hall/CRC. https://doi.org/10.1201/9781420064254 Li, C.S. and Liu, Z.J. 2012. Study on the relationship among Chinese unemployment rate, economic growth and inflation. Advance in Applied Economics and Finance 1. (1): 1–6. Louail, B. and Riache, S. 2019. Asymmetry relation- ship between economic growth and unemployment rates in the Saudi economy: Application of Okun’s law during the period. International Journal of Advanced and Applied Sciences 6. (10): 83–88. https:// doi.org/10.2478/manment-2019-0070 Madito, O. and Khumalo, J. 2014. Economic growth- unemployment nexus in South Africa: VECM approach. Mediterranean Journal of Social Sciences 5. (20): 79–84. https://doi.org/10.5901/mjss.2014. v5n20p79 Makaringe, S.C. and Khobai, H. 2018. The ef- fect of unemployment on economic growth in South Africa (1994–2016). Working Papers 1815. Gqeberha, South Africa, Department of Economics, Nelson Mandela University. https://mpra.ub.uni- muenchen.de/85305/1/MPRA_paper_85305.pdf Maza, A. 2022. Regional differences in Okun’s law and explanatory factors: Some insights from Europe. International Regional Science Review 45. (5): 555–580. https://doi.org/10.1177/01600176221082309 Montero Kuscevic, C.M. 2014. Okun’s law and urban spillovers in US unemployment. The Annals of Regional Science 53. (3): 719–730. https://doi. org/10.1007/s00168-014-0640-2 Moran, P.A.P. 1948. The interpretation of statistical maps. Journal of the Royal Statistical Society Series B 10. 243–251. https://doi.org/10.1111/j.2517-6161.1948. tb00012.x Nikolli, E. 2014. Economic growth and unemploy- ment rate. Case of Albania. European Journal of Social Science Education and Research 1. (1): 217–227. https:// doi.org/10.26417/ejser.v1i1.p217-227 Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176. https://doi.org/10.5901/mjss.2015.v6n4s3p454 https://doi.org/10.5901/mjss.2015.v6n4s3p454 https://doi.org/10.1007/978-3-662-05617-2 https://doi.org/10.1007/978-3-662-05617-2 https://doi.org/10.1111/jmcb.12420 https://doi.org/10.1007/s12076-017-0189-7 https://doi.org/10.1007/978-3-642-03647-7_19 https://doi.org/10.1016/j.regsciurbeco.2021.103756 https://doi.org/10.1016/j.regsciurbeco.2021.103756 https://doi.org/10.1515/cks-2016-0007 https://doi.org/10.1515/cks-2016-0007 https://doi.org/10.2307/2953682 https://doi.org/10.2307/2953682 https://doi.org/10.3390/en14185829 https://doi.org/10.3390/en14185829 https://doi.org/10.15201/hungeobull.72.2.6 https://doi.org/10.2307/2060063 https://doi.org/10.1201/9781420064254 https://doi.org/10.2478/manment-2019-0070 https://doi.org/10.2478/manment-2019-0070 https://doi.org/10.5901/mjss.2014.v5n20p79 https://doi.org/10.5901/mjss.2014.v5n20p79 https://ideas.repec.org/s/mnd/wpaper.html https://mpra.ub.uni-muenchen.de/85305/1/MPRA_paper_85305.pdf https://mpra.ub.uni-muenchen.de/85305/1/MPRA_paper_85305.pdf https://doi.org/10.1177/01600176221082309 https://doi.org/10.1007/s00168-014-0640-2 https://doi.org/10.1007/s00168-014-0640-2 https://doi.org/10.1111/j.2517-6161.1948.tb00012.x https://doi.org/10.1111/j.2517-6161.1948.tb00012.x https://doi.org/10.26417/ejser.v1i1.p217-227 https://doi.org/10.26417/ejser.v1i1.p217-227 Jankiewicz, M. Hungarian Geographical Bulletin 74 (2025) (2) 163–176.176 Oberst, C. and Oelgemöller, J. 2013. Economic growth and regional labour market development in German regions: Okun’s law in a spatial context. FCN Working Paper 5. (2013). E.ON Energy Research Centre. https://doi.org/10.2139/ssrn.2240260 Okoro-Ugochukwu, N.A. and Adenomon, M.O. 2021. Application of Okun’s law on Nigerian economy: Evidence from spatial econometrics analysis. Science World Journal 16. (3): 291–295. Okun, A.M. 1962. Potential GNP: Its measurement and significance. In Proceedings of the Business and Economics Section. Boston, American Statistical Association, 98–103. Onakoya, A.B. and Seyingbo, A.V. 2020. Economic growth and unemployment nexus: Okun’s two- version case for Nigeria, South Africa, and United States of America. Journal of Economics and Behavioral Studies 12. (1): 55–65. https://doi.org/10.22610/jebs. v12i1(J).3006 Palombi, S., Perman, R. and Tavéra, C. 2015. Regional growth and unemployment in the medium run: Asymmetric co-integrated Okun’s law for UK re- gions. Applied Economics 47. (57): 6228–6238. https:// doi.org/10.1080/00036846.2015.1068922 Palombi, S., Perman, R. and Tavéra, C. 2017. Commuting effects in Okun’s law among British areas: Evidence from spatial panel econometrics. Papers in Regional Science 96. (1): 191–209. https:// doi.org/10.1111/pirs.12166 Pereira, R.M. 2014. Okun’s law, asymmetries and regional spillovers: Evidence from Virginia metropolitan statistical areas and the District of Columbia. The Annals of Regional Science 52. (2): 583–595. https://doi.org/10.1007/s00168-014-0600-x Sadiku, M., Ibrahimi, A. and Sadiku, L. 2015. Econometric estimation of the relationship between unemployment rate and economic growth of FYR of Macedonia. Procedia Economics and Finance 19. 69– 81. https://doi.org/10.1016/S2212-5671(15)00009-X Salvati, L. 2015. Space matters: Reconstructing a local-scale Okun’s law for Italy. International Journal of Finance, Insurance and Risk Management 5. (1): 833–840. Schabenberger, O. and Gotway, C.A. 2005. Statistical Methods for Spatial Data Analysis. Boca Raton– London–New York, Champion & Hall/CRC. Soylu, Ö.B., Çakmak, İ. and Okur, F. 2018. Economic growth and unemployment issue: Panel data analysis in Eastern European countries. Journal of International Studies 11. (1): 93–107. https://doi. org/10.14254/2071-8330.2018/11-1/7 Szulc, E. and Jankiewicz, M. 2018. Spatio-temporal modelling of the influence of the number of busi- ness entities in selected urban centres on unemploy- ment in the Kujawsko-Pomorskie voivodeship. Acta Universitatis Lodziensis. Folia Oeconomica 4. (337): 21–37. https://doi.org/10.18778/0208-6018.337.02 Tobler, W.R. 1970. A computer movie simulating urban growth in the Detroit region. Economic Geography 46. Supplement 1. 234–240. https://doi. org/10.2307/143141 Valadkhani, A. and Smyth, R. 2015. Switching and asymmetric behaviour of the Okun coefficient in the US: Evidence for the 1948–2015 period. Economic Modelling 50. 281–290. https://doi.org/10.1016/j. econmod.2015.07.001 Vega, S.H. and Elhorst, J.P. 2013. On spatial econometric models, spillover effects, and W. In 53rd ERSA Congress “Regional Integration: Europe, the Mediterranean and the World Economy”, 27–31 August 2013. Palermo, Italy. Louvain-la-Neuve, ERSA, 1–13. Villaverde, J. and Maza, A. 2016. Okun’s law among Spanish regions: A spatial panel approach. XLI Reunión de Estudios Regionales, Reus. Xu, B., Yu, H. and Li, L. 2021. The impact of entre- preneurship on regional economic growth: A per- spective of spatial heterogeneity. Entrepreneurship & Regional Development 33. (3–4): 309–331. https:// doi.org/10.1080/08985626.2021.1872940 https://doi.org/10.2139/ssrn.2240260 https://doi.org/10.22610/jebs.v12i1(J).3006 https://doi.org/10.22610/jebs.v12i1(J).3006 https://doi.org/10.1080/00036846.2015.1068922 https://doi.org/10.1080/00036846.2015.1068922 https://doi.org/10.1111/pirs.12166 https://doi.org/10.1111/pirs.12166 https://doi.org/10.1007/s00168-014-0600-x https://doi.org/10.1016/S2212-5671(15)00009-X https://doi.org/10.14254/2071-8330.2018/11-1/7 https://doi.org/10.14254/2071-8330.2018/11-1/7 https://doi.org/10.18778/0208-6018.337.02 https://doi.org/10.2307/143141 https://doi.org/10.2307/143141 https://doi.org/10.1016/j.econmod.2015.07.001 https://doi.org/10.1016/j.econmod.2015.07.001 https://doi.org/10.1080/08985626.2021.1872940 https://doi.org/10.1080/08985626.2021.1872940 Mateusz JANKIEWICZ: Analysis of spatial dependencies and spatial effects in the relationship between economic growth and unemployment in Europe. Hungarian Geographical Bulletin Vol 74 Issue 2 163-176