Wage differences in poland at the county level and their determinants

Luśtyk, Agata; Połeć, Anna; Voznyuk, Inna Article Wage differences in poland at the county level and their determinants Central European Economic Journal (CEEJ) Provided in Cooperation with: Faculty of Economic Sciences, University of Warsaw Suggested Citation: Luśtyk, Agata; Połeć, Anna; Voznyuk, Inna (2024) : Wage differences in poland at the county level and their determinants, Central European Economic Journal (CEEJ), ISSN 2543-6821, Sciendo, Warsaw, Vol. 11, Iss. 58, pp. 447-460, https://doi.org/10.2478/ceej-2024-0028 This Version is available at: https://hdl.handle.net/10419/324625 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ ISSN: 2543-6821 (online) Journal homepage: http://ceej.wne.uw.edu.pl To cite this article Luśtyk, A., Połeć, A., Voznyuk, I. (2024). Wage Differences in Poland at the County Level and their Determinants. Central European Economic Journal, 11(58), 447-460. DOI: 10.2478/ceej-2024-0028 To link to this article: https://doi.org/10.2478/ceej-2024-0028 Wage Differences in Poland at the County Level and their Determinants Agata Luśtyk, Anna Połeć, Inna Voznyuk Open Access. © 2024 A. Luśtyk, A. Połeć, I. Voznyuk, published by Sciendo. This work is licensed under the Creative Commons Attribution 4.0 International License. Agata Luśtyk Jagiellonian University in Kraków, Doctoral School in the Social Sciences, Rynek Główny 34, 31-010 Cracow, Poland corresponding author: [email protected] Anna Połeć Jagiellonian University in Kraków, Doctoral School in the Social Sciences, Rynek Główny 34, 31-010 Cracow, Poland Inna Voznyuk Jagiellonian University in Kraków, Faculty of International and Political Studies, Władysława Reymonta 4, 30-059 Cracow 1. Introduction The political transformation in Poland after 1989 entailed dramatic changes in various sectors of its economy, thus greatly influencing the labour market structure. Privatisation, the formation of profit-making businesses, and mass layoffs led to the unemployment rate increasing to more than 10% between 1991 and 2007. The privatisation processes also contributed to an inflow of foreign capital to Poland, posing a major challenge to Polish businesses and forcing them to adopt more efficient operation methods. This resulted in an increase in labour productivity with time (Adamowicz, 2022). Labour productivity rose faster than wages between 1995 and 2008. Competition and changes in the sectoral structure of wages had a marginal effect on the remuneration level, and total employment fell from about 4.3 million workers in 1995 to 3.4 million in 2002 in a group of enterprises surveyed by Growiec (2009). Wage Differences in Poland at the County Level and their Determinants Abstract This study investigates the impact of unemployment and labour productivity on relative wages in Polish counties (powiats) from 2008 to 2021. Labour productivity is measured as the ratio of sold industrial production to the number of workers. The data is sourced from the Local Data Bank of Statistics Poland. The analysis employs the Solow model of efficiency wages, the neoclassical Solow model, and the Durbin model of spatial econometrics. The results reveal that both unemployment and labour productivity are statistically significant in explaining relative wages, with unemployment having the strongest, albeit negative, effect during the study period. Notably, changes in unemployment rates or wages in a county influence wage changes in neighbouring counties. The issue of spatial wage differences at the county level in Poland has not been sufficiently explored in recent years. Although recent research has focused on regional (voivodeship-level) wage differences, there remains a gap in understanding wage differences at the county level. Given changes in the Polish labour market, particularly due to the COVID-19 pandemic, this study aims to update previous findings and provide a more detailed analysis. Keywords labour market indicators | Solow model of efficiency wages | neoclassical Solow model | spatial Durbin model JEL Codes E24, J31, J40 CEEJ • 11(58) • 2024 • pp. 447-460 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0028 449 After the economic slowdown in 2001-2002 caused by a fall in economic growth in Western Europe and a slump in exports to eastern markets, conditions in the labour market improved in 2006 with both supply and demand. A moderate but perceptible annual reduction in the unemployment rate was also observed: from 13.9% in 2006 to 3.3% in 2019 among those aged 15-74 (Kowalik, Magda, 2021). Poland’s accession to the European Union expanded the emigration of Poles to other EU member states, leading to an increase in demand for workforce in selected sectors of the Polish economy. The aim of this study is to provide an empirical analysis of wage determinants by county between 2008 and 2021. It also presents an analysis of spatial and temporal variations in unemployment rates, labour productivity and wages. The main study uses the spatial Durbin model to estimate the parameters in the equation that describes relative wages. The two parameters are the unemployment rate and labour productivity, computed as the ratio of sold production of industry to the number of workers by county and year. The data for empirical analysis are obtained from the Local Data Bank of Statistics Poland. The analysed time interval represents the longest period for which information is available about wages, the unemployment rate, sold production of industry and the number of workers. Importantly, sold production of industry is an optimum substitute for Gross Domestic Product, whose values are unavailable at the county level. The values of the variables analysed are expressed at constant prices from 2021, converted using annual inflation rates published by Statistics Poland. The study demonstrates which of the determinants computed by county over the period under study influence wages, the nature of their influence, and their significance levels. The results obtained can undoubtedly assist local authorities in defining directions for labour market development, and particularly lead to an increase in wage indices at the regional level. This article is structured as follows. The initial section contains a review of the literature on the subject of research. The second section presents an analysis of spatial and temporal differences in selected independent variables and in the dependent variable. The authors then present the theory underlying the Solow model of efficiency wages. The last section focuses on presenting the model’s results and conclusions drawn from the analysis. 2. Literature review The literature on the subject proposes numerous analyses of spatial wage differentials at the county level (Adamczyk et al., 2009; Dykas & Misiak, 2014; Przekota, 2016), voivodeships (Kapela & Kwiatkowski, 2023) and the Ukrainian oblast (region) level (Bolińska & Gomółka, 2018; Dykas et al., 2020). As regards the theoretical approach, the Solow (1979) and Summers (1988) models of efficiency wages is frequently used, in combination with neoclassical long-run economic growth models as proposed by Solow (1956), Mankiw, Romer and Weil (1992) or Nonneman and Vanhoudt (1996). As to the empirical content, statistical analysis includes such basic indicators as wages, labour productivity or unemployment rates; theoretical models are then used to estimate the parameters in the regression equation wherein wages are explained by labour productivity and the unemployment rate. Wages are also explained by such variables as the working age population, capital expenditures, sold production of industry, the number of businesses per 1,000 residents or the number of training programmes. Kapela and Kwiatkowski (2023) in their latest study used a more detailed set of explanatory variables for average gross wages in the voivodeships. In addition to productivity and the average unemployment rate in the voivodeship, they used such variables as the share of people with higher education in total employment in the voivodeship, the share of the value of sales of products of entities classified as high and medium-high technology in the net income from sales of products of entities classified as industry, the share of the number of innovative enterprises in the total number of enterprises in the voivodeship, and the number of patents granted by the Patent Office of the Republic of Poland per 100,000 residents in the voivodeship. The authors also included the effect of the 2020 pandemic in their analysis. Combes, Duranton and Gobillon (2008) presented an interesting approach to explaining the occurrence of large spatial wage differences among French workers. They distinguished three aspects of the incidence of wage differences: the skills of the workforce, equipment (specifically local non-human resources, i.e. favourable location, climate better suited to doing business, local institutions or technologies) and intraand inter-industry interactions. Considering methods used to estimate the parameters in the regression equation, the method of least squares prevails, but the generalized method of CEEJ • 11(58) • 2024 • pp. 447-460 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0028 450 moments is also used. Additionally, there are analyses that use simple rank-order methods, coefficients of variation, other clustering methods and, of course, econometric models with fixed and random effects. A fixed effects model is also used in analyses at a local level. This approach generally results in an improved goodness-of-fit index. One of the main conclusions concerning research on the relationship between the unemployment rate and real wages posits a negative wage elasticity relative to the unemployment rate (Bolińska & Gomółka, 2018). Research conducted at the county level also demonstrates that higher wages are characteristic of counties located near big cities and of urban counties, while the lowest wages are characteristic of agricultural counties. Further, an increase in a county’s labour productivity entails an increase in relative wages (Adamczyk et al., 2009). Past analyses also confirm a positive effect of desirable changes in the working age population on the rate of wage growth and a negative effect of the rate of increase in the number of businesses on the rate of wage growth. Studies of the relationship between the rate of changes in capital expenditures, the rate of changes in sold production of industry and average wages at the county level produced ambiguous results (Przekota, 2016). Additionally, research on wages in the Iranian economy shows that an increase in the number of training programmes leads to a rise in workforce wage rates (Karrari & Ersungur, 2019). Interestingly, research on counties using a fixed effects model demonstrates a positive relationship between wages and the unemployment rate (Dykas & Misiak, 2014). The latest results of analysis at the voivodeships level covering the period 2010-2020 show that the wage gap between voivodeships narrowed slightly during this period. The econometric analysis confirmed that the share of people with higher education, the number of patents and the share of innovative enterprises also had a positive impact on the level of wages, in addition to the previously mentioned variables. The pandemic in 2020 also had a significant positive impact on the level of wages in the voivodeships (Kapela & Kwiatkowski, 2023). Finally, the study also revealed that differences in the composition of labour force skills account for 40-50% of aggregate spatial wage differences. In addition, the role of interaction explanations (in particular, the variable related to population density, i.e. urbanisation) is moderate, while the role of equipment is weak in explaining spatial wage differences (Combes et al., 2008). Alban W. Phillips indicated unclear relations between wages and the unemployment rate as early as 1958. In his study of data on the British labour market, for example, he demonstrated a negative correlation between the unemployment rate and the rate of money wage growth. He considered the rate of increase in unemployment to be a major factor, indicating that the direction of changes in unemployment may affect the direction of changes in wages (Phillips, 1958). However, the most important conclusion drawn from his study was a curve showing changes in wages as a decreasing and convex function of unemployment. Stephan F. Kaliski drew from the studies by Phillips and Lipsey when studying an inverse relationship between wages and unemployment, increase in unemployment or increase in average price level in his case study of Canada, arriving at similar conclusions. Empirical analysis also showed a decreasing negative effect of unemployment on wages in the late 20th century, and an increasing correlation with the price level (Kaliski, 1964). More recently, some authors’ analyses of the wage curve have indicated a specific unemployment level above which the elasticity of wages relative to unemployment was no longer negative. Moreover, the effect of unemployment on wages above that level was no longer statistically significant or became positive (Blanchflower & Oswald, 1990). Opposite results were produced by studies conducted in South Africa where unemployment rates reached high values, but there was a negative elasticity of wages relative to the unemployment rate (Kingdon & Knight, 2006). References to the Phillips curve can also be found in current studies. Bartosik and Mycielski, using the wage curve model proposed by Phillips, examined the relationship between money wages and longterm unemployment in the Polish economy. Machuca and Cota used the same model to study the Mexican economy and assess the relationship between real wages, labour productivity and unemployment. Their study demonstrated an inverse relationship between wages and labour productivity and unemployment (Bartosik & Mycielski, 2015; Machuca & Cota, 2017). Another important topic discussed in the literature is that of real wages, considering the distinct influence of the price level on money (or nominal) wages. An example is a comparison of wages in the North and South of the United States in years 1973-1978. The historical trend in money wages was slightly reversed, and the real wage differential showed a strong trend toward reversal (Sahling & Smith, 1983). The problem of differences in determinants and their impact on relative wages by county in recent CEEJ • 11(58) • 2024 • pp. 447-460 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0028 451 years has not been fully explored. The last similar research, using the same variables, was conducted by Dykas and Misiak in 2014 and covered the period 2002-2011. In contrast, the most recent research focusing on the analysis of spatial wage differences, which arose in parallel with the present one, focuses primarily on the voivodeships level and takes into account a wider range of variables. Therefore, the need arises to address this problem and have results updated at the county level. A similar study of relative wages in Poland at the county level was conducted by Adamczyk, Tokarski and Włodarczyk in 2009 and covered the period 2002-2006. While many general conclusions prevail, some aspects of the labour market have changed since then. Most importantly, the present study examines the impact of the COVID-19 pandemic on the Polish labour market. The present paper also includes additional findings of change patterns over time for three groups of counties. These groups were chosen based on their economic history and current funding programmes. Additionally, spatial effects are discussed. Considering the outlined research problem, the following hypotheses are proposed: H1: The unemployment rate had the strongest influence on relative wages in Poland at the county level in 2008-2021. H2: Changes in unemployment in a county between 2008 and 2021 affected changes in relative wages in adjacent counties. H3: Changes in relative wages in a county between 2008 and 2021 affected changes in relative wages in adjacent counties. H4: The unemployment rate and relative labour productivity statistically significantly explained relative wage rates by county between 2008 and 2021. 3. Spatial and temporal differences in unemployment rates, labour productivity and wages In examining differences in unemployment rates, labour productivity and gross wages by county in the years 2008-2021, both the spatial and the temporal approach was adopted. Based on the data, the value of an indicator was computed for each county as a mean of values recorded year by year. These results were used to make an analysis of spatial differentials. In addition, counties were divided into three groups, namely Eastern Poland counties, Western Poland counties and Central Poland counties. Given this division, we computed the arithmetic mean, for each indicator in each year, of the values recorded in the counties falling into a category. These results were used to make a supplementary spatial analysis with an additional temporal analysis. A separate analysis was also made, concerning urban counties against the background of other counties. Vast differences were revealed in the average unemployment rate in 2008-2021. The average unemployment rate from all counties amounted to about 11.97% with standard deviation around 5.02%, which means that the coefficient of variation was about 41.93%. The lowest values were characteristic of urban counties and their adjacent counties. This type of counties comprises almost half of counties falling into the first quintile. In the city of Poznań, the average unemployment rate was about 2.52%, and in Poznań County, 2.67%. In the capital city of Warsaw, the average unemployment rate was 2.81%. The highest unemployment rates were recorded in selected counties of Eastern and Central Poland. In Pisz, Łobez, Białogard, Radom and Kętrzyn Counties, it exceeded 24%. In Bartoszyce and Braniewo Counties (Warmian–Masurian Voivodeship), the average unemployment rates were 25.35% and 26.46%, respectively. The highest unemployment rate was recorded in Szydłowiec County (Masovian Voivodeship), amounting to 30.91%. In the group of eight counties where the average unemployment rate exceeded 24%, half were counties located in Eastern Poland. However, the distribution of county groups in the last quintile was quite uniform, although counties located in Eastern Poland still represented the largest percentage, precisely 38.16%. Spatial differences in unemployment are shown on Map 1. There is a characteristic that cannot be ignored here: each macroregional group includes counties with an unemployment rate exceeding 20%, and this makes the results (with the exception of the distinctly lower values among urban counties) relatively comparable, although worse rates are observed in Eastern Poland. Indeed, the average unemployment rate over the entire period in Western and Central Poland amounted to 11.45% and 11.19%, respectively. In Eastern Poland, CEEJ • 11(58) • 2024 • pp. 447-460 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0028 452 it reached 13.85%. This is also confirmed by a study of temporal changes in unemployment rates characteristic of the indicated county groups as shown on Chart 1. It demonstrates that even if the change pattern of the unemployment rate in Eastern Poland counties was similar to that observed in the other two groups, the rate was higher over the entire period by about 2.5 percentage points. The change pattern between 2009 and 2013 indicated a general rise in unemployment. The highest unemployment levels were also recorded in those years. A minor deviation from the general trend was observed in Western Poland counties where the unemployment rate slightly fell in 2009-2011. In 20092013, the average unemployment level in Eastern Poland counties exceeded 16%. At its peak moment in 2013, it reached 18.18%. However, a clear downward trend was observed in 2013-2019. In 2019, the average unemployment rate in urban counties was 4.5%. This statistic also had comparable values in Central and Western Poland, where the average was 6.21%. As of 2013, an absolute difference between unemployment rates in these county groups was less than 0.5 p.p. The year 2020 marked a deviation from the trend, possibly caused by the COVID-19 pandemic, but the indicator began to fall in Western and Central Poland again in 2021. In Western Poland, it reached a level comparable to that from 2018. However, an increase in unemployment in Eastern Poland was observed as of 2020. Labour productivity, computed as the ratio of sold production (expressed in PLN million, base year Map 1. Spatial differences in the unemployment rate by county in 2008-2021 [%] Source: Own elaboration CEEJ • 11(58) • 2024 • pp. 447-460 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0028 453 2021) to the number of workers, did not exceed PLN 0.4 million per worker on average. The average from all counties during the studied period amounted to about PLN 0.099 million per worker. The standard deviation was about PLN 0.072 million per worker, which translates to a 72.87% coefficient of variation. A notable exception is the city Dąbrowa Górnicza, where average labour productivity in 2008-2021 was about PLN 0.57 million per worker. The group with high labour productivity is dominated by counties from Silesian Voivodeship and selected counties from Lower Silesian Voivodeship, plus Bełchatów County in Łódź Voivodeship. In Silesian Voivodeship, in addition to Dąbrowa Górnicza, the highest labour productivity was recorded in Bielsko-Biała town, where it reached about PLN 0.394 million per worker. Generally, labour productivity reached a relatively high level in multiple towns of the Silesian urban area. It amounted to about PLN 0.26 million per worker in Gliwice, to PLN 0.24 million per worker in Siemanowice Śląskie, and to PLN 0.22 million per worker in Jastrzębie-Zdrój. A clearer trend could be observed in counties with low labour productivity. Eastern Poland counties dominated this group. From among 38 counties where labour productivity was contained in the interval between PLN 0.005 million and 0.03 million per worker, 26 were located in Eastern Poland. The relationships described are presented on Map 2. Average results for the county groups over the entire period are also interesting. The highest labour productivity was observed in Western Poland, amounting to PLN 0.12 million per worker on average. In Central Poland, average labour productivity equalled about PLN 0.108 million per worker. In Eastern Poland, it was PLN 0.06 million per worker. A vast difference also occurred between urban and rural counties. Average labour productivity in urban counties was PLN 0.122 million per worker, while rural counties reached PLN 0.0987 million per worker. Temporal differences in labour productivity per county group presented on Chart 2 clearly indicate that, except for a slight decline in Western Poland in 2008-2010, labour productivity in each county group showed a rising trend (or remained at a fairly constant level). However, a difference in labour productivity in absolute terms existed between the county groups over the entire period under study. The lowest level of labour productivity was characteristic of Eastern Poland. In 2010, it dropped to the lowest value of PLN 0.049 million per worker. However, like the other groups, it showed a general upward trend over the period under study, although its rise was rather slow prior to 2019, with the value of labour productivity in 2011-2016 remaining at a near constant level. Moreover, the difference between labour productivity in Eastern Poland and its average level in Western and Central Poland gradually rose from about PLN 0.04 million per worker to almost PLN 0.06 million per worker in 2019. Additionally, Western and Central Poland counties achieved a strong rise in 2020 and Chart 1. Change pattern of the unemployment rate by macroregions in 2008-2021 [%] Source: Own elaboration CEEJ • 11(58) • 2024 • pp. 447-460 • ISSN 2543-6821 • DOI: 10.2478/ceej-2024-0028 454 Map 2. Spatial differences in labour productivity by county in 2008-2021 [PLN million/worker, constant prices from 2021] Source: Own elaboration Chart 2. Change pattern of labour productivity by macroregions in 2008-2021 [PLN million/worker, constant prices from 2021] Source: Own elaboration