Abstract
Republic of Poland: Selected Issues
The Role of Productivity Growth in Reducing Regional Economic Disparities in Poland1
Although Poland has enjoyed strong growth and steady income convergence with the EU over the last two decades, important disparities persist at the regional level. Per-capita income is higher in the west—which is integrated into the German supply chain and enjoys higher levels of FDI—than in the east—where the economy depends more on less productive agriculture. Despite strong overall economic growth, the east has not been catching up to the west. This chapter identifies policies to increase productivity in the east, reduce regional income disparities, and promote overall income convergence. This would require improving educational attainment and reducing skill mismatches in the east, scaling up public infrastructure to attract investment to less productive regions, and facilitating labor mobility.
A. Regional Disparities and Income Convergence
1. Poland has enjoyed strong economic performance over the last two decades. Economic convergence to the EU has progressed steadily, with Poland closing over a quarter of its per capita income gap with the EU28 average during 1995–2014. Prudent economic policies that had prevented the build-up of imbalances in the run-up to the global financial crisis helped Poland avoid an outright recession during the crisis; GDP growth has averaged around 3 percent since 2008. At the same time, overall income inequality has declined to the EU average.
GDP Per Capita Relative to the EU
(Percent of EU28 average, PPS per capita)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Eurostat.GDP Per Capita Relative to the EU
(Percent of EU28 average, PPS per capita)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Eurostat.GDP Per Capita Relative to the EU
(Percent of EU28 average, PPS per capita)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Eurostat.
Income Inequality, 2005-14
(Gini index of equivalized disposable income)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Eurostat.Income Inequality, 2005-14
(Gini index of equivalized disposable income)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Eurostat.Income Inequality, 2005-14
(Gini index of equivalized disposable income)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Eurostat.2. However, this strong overall performance masks important economic disparities at the regional level. Western regions of the country are wealthier than eastern regions: on average GDP per capita is about 30 percent higher in regions bordering Germany in the west, than in regions bordering Belarus and Ukraine in the east. The wealthiest region, Mazowieckie (which includes the capital city Warsaw), has a per capita income comparable to France—the third largest economy in the EU, while the poorest regions on the eastern border have per capita incomes comparable to Bulgaria—the poorest in the EU by income level. Poverty rates and long-term unemployment are also higher in the east.
3. These disparities have been persistent. Poland’s regions have been growing at about the same rate during the past decade, making it difficult for poorer regions to catch up with wealthier regions (Figure 1).2
Poland: Regional Economic Disparities
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Central Statistical Office, Eurostat, and IMF staff calculations.Poland: Regional Economic Disparities
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Central Statistical Office, Eurostat, and IMF staff calculations.Poland: Regional Economic Disparities
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Central Statistical Office, Eurostat, and IMF staff calculations.4. Reducing regional disparities by boosting economic performance in lagging regions would promote faster and more inclusive growth. Estimates suggest that Poland’s GDP per capita could be increased by 7 percent if one-third of the gap between regions with GDP per capita below average and the four regions which are above average was closed. This would imply an additional 5 percentage point convergence to the EU28 average income level.
5. Boosting labor productivity growth in lagging regions is key to reducing regional disparities. There is a strong association between income per capita and labor productivity levels: wealthier regions in Poland also tend to have higher levels of productivity.3
GDP per Capita and Labor Productivity by Region, 2012
(Percent of country average)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Central Statistical Office, Eurostat, and IMF staff calculations.GDP per Capita and Labor Productivity by Region, 2012
(Percent of country average)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Central Statistical Office, Eurostat, and IMF staff calculations.GDP per Capita and Labor Productivity by Region, 2012
(Percent of country average)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Central Statistical Office, Eurostat, and IMF staff calculations.6. This chapter looks at determinants of regional productivity growth in Poland.
Section B establishes stylized facts about potential drivers of regional productivity differences. Section C describes the empirical model and discusses the key statistical determinants of labor productivity growth. Section D summarizes the findings and sets out policy recommendations.
B. Regional Productivity in Poland: Stylized Facts
7. Certain economic variables appear correlated with regional productivity levels. These factors are highlighted below and illustrated in Figure 2:
Poland: Structural Indicators and Labor Productivity
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Central Statistical Office, Eurostat, and IMF staff calculations.Poland: Structural Indicators and Labor Productivity
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Central Statistical Office, Eurostat, and IMF staff calculations.Poland: Structural Indicators and Labor Productivity
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Central Statistical Office, Eurostat, and IMF staff calculations.Regions with higher foreign direct investment (FDI) have higher productivity. A higher intensity of FDI in more productive regions suggests that foreign investment has boosted productivity by bringing in technological and organizational know-how. Poland’s integration into the German supply chain explains a higher share of FDI in the western regions bordering Germany.
Educational attainment is correlated with regional productivity. Skilled labor is likely to be more productive, and regions with easier access to skilled workers may attract more technologically advanced, productive businesses.
Regions with low productivity have a higher share of agriculture. Agriculture is the least productive sector of the economy, contributing to the negative correlation between regional productivity and employment in agriculture. Regions in the west on average have larger farms and a lower share of employment in agriculture than regions in the east. As discussed later, there is potential to boost productivity by reallocating labor from agriculture to other sectors.
EU funds support less productive regions. Funds available under the EU cohesion policy are aimed at helping the development of poorer regions. Spending of such funds in Poland is either decided centrally (for example on highways) or locally under regional programs. The latter have been directed more to less productive regions in eastern Poland.
More productive regions have a better transportation network. Transportation network is an important factor affecting location of business, especially in exporting industries that are part of international supply chains. The current network of highways is more likely to attract business to the more productive western regions. The transportation network in Figure 2 shows that major highways link Poland’s western, central, and southern regions to Germany. The density of railroads is also higher in western regions, facilitating the movement of goods and improving labor mobility.
Unemployment is higher in less productive regions. While unemployment rates vary a lot among regions, higher unemployment rates are observed in less productive regions of Poland. Since unemployment incidence is higher among less educated persons, in some regions there may be a vicious circle of weak regional educational attainment, high unemployment, and low productivity.
8. Wealthier regions have higher productivity across all sectors. Data on regional productivity by economic sector suggest that the comparative advantage of richer regions is not due to a particular type of economic activity. In general, wealthier regions tend to display higher productivity across most sectors.
Labor Productivity by Region and Sector, 2012
(Thousand PLN per employed)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Central Statistical Office, Eurostat, and IMF staff calculations.Labor Productivity by Region and Sector, 2012
(Thousand PLN per employed)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Central Statistical Office, Eurostat, and IMF staff calculations.Labor Productivity by Region and Sector, 2012
(Thousand PLN per employed)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Central Statistical Office, Eurostat, and IMF staff calculations.C. Empirical Analysis
9. This section conducts a more rigorous empirical analysis of the factors determining labor productivity growth at the regional level (Box 1). Determinants of productivity growth are chosen to reflect structural factors and spillover effects. The following structural variables are considered:
Labor quality (share of population with education below secondary school). Better educated workers are more likely to innovate, quickly adapt to changing technology and relocate towards more productive firms. Access to a well-educated workforce is also an important pull factor for new investments.4
Structural transformation (change in the share of employment in agriculture relative to industry and services). Productivity growth patterns are dependent on regional economic structure, whereby productivity growth in a given region is affected by reallocation of labor from low- to high-productivity sectors through the process of structural transformation (Box 2). Agriculture is the least productive sector in the economy, and the analysis controls for changes in the structure of the economy by including the change in the share of employment in agriculture relative to other sectors. In addition, to account for the impact of labor reallocation within industry and services, changes in the shares of employment in these sectors are also considered.
Poland: Regions
Poland is divided into 16 administrative regions (voivods), 380 counties (powiats) and 2,479 municipalities (gminas). On average, the population of a voivod is 2.4 million persons, and is close to 100,000 persons for a powiat.
The European Union Nomenclature of Territorial Units (NUTS) classification is used in this analysis. Within the NUTS classification, NUTS level 2 includes the 16 administrative voivods, and NUTS level 3 consists of 66 regions, which are groupings of powiats. Regions excluded from the analysis are Warsaw, the capital city, Ciechanowsko-Plocki, which hosts the largest oil refinery (Orlen), and Legnicko-Glogowski which includes the largest copper company (KGHM).
Statistical division of Poland
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: EurostatStatistical division of Poland
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: EurostatStatistical division of Poland
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: EurostatUnemployment. Productivity growth depends on firms’ choice of input factors composition. Structural unemployment may differ across regions due to heterogeneity in search costs, quality of labor and labor mobility. High structural unemployment could also result from low labor mobility or inability of workers to train and change sectors, creating over-supply and low productivity in some sectors. Additionally, weaker labor markets with underutilized or idle workers reduce employee bargaining power, and encourage substitution of capital with labor. Conversely, high search costs increase the marginal cost of labor and could encourage substitution of labor with capital.
Ability to innovate. Regions with more innovative enterprises should display faster productivity growth, if innovations are applied locally and support more efficient use of resources. As a proxy, the number of patent applications to the European Patent Organization per million employed persons is used.
Local government finances. Local government expenditures account for one-third of total general government spending. As such, the composition and size of local budgets may have an impact on productivity growth. The share of investment spending in total expenditures is used to explore whether local investments boost productivity. Nonetheless, this is an imperfect measure because it does not capture possibly important investments financed by the central government.5 The share of non-investment spending to gross value added is also used to control for the impact of local government size on productivity.
Poland: The Role of Structural Transformation and Cross-Regional Reallocation of Labor in Productivity Growth
Reallocation of labor across sectors of the economy affects productivity. The overall change in productivity can be decomposed into two parts: (i) within-sector changes and (ii) changes due to reallocation of labor across sectors (see Appendix III). The latter process, called structural transformation, supported productivity growth in Poland in the past, and its impact was larger than in regional peer countries.1
Contribution of Labor Reallocation across Sectors to Productivity Growth, 2000-12
(Percentage Points)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Central Statistical Office, Eurostat, and IMF staff calculations .Contribution of Labor Reallocation across Sectors to Productivity Growth, 2000-12
(Percentage Points)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Central Statistical Office, Eurostat, and IMF staff calculations .Contribution of Labor Reallocation across Sectors to Productivity Growth, 2000-12
(Percentage Points)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Central Statistical Office, Eurostat, and IMF staff calculations .Structural transformation added more to productivity gains in poorer regions. If regional productivity growth is decomposed as mentioned above, a remarkable difference arises between the east and west of Poland. Poorer regions in the east benefitted visibly more from structural transformation, which likely reflects their higher share of agriculture, implying greater opportunity to shift labor from low-productivity farming to more productive sectors. This also may help to explain why such regions kept productivity growth apace with their richer peers.
Cross-Regional Reallocation of Labor Contribution to Productivity Growth, 2000-12 (pp)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Central Statistical Office, Eurostat, and IMF staff calculations.Cross-Regional Reallocation of Labor Contribution to Productivity Growth, 2000-12 (pp)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Central Statistical Office, Eurostat, and IMF staff calculations.Cross-Regional Reallocation of Labor Contribution to Productivity Growth, 2000-12 (pp)
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Source: Central Statistical Office, Eurostat, and IMF staff calculations.Bottlenecks appear to exist in cross-regional mobility of labor. Productivity is also affected by reallocation of labor across regions.2 At the aggregate level, such a process played a rather limited role, contributing just 2.5 percentage points to total productivity gains in 2000–12. Sector analysis of productivity gains from reallocation of labor among regions but within the same sector shows that only industry and services benefitted from such cross-regional labor movements. Such a picture suggests bottlenecks in cross-regional labor mobility. It also may signal adverse policy incentives in sectors like agriculture, discouraging reallocation of labor to more productive sectors and regions.
1 A detailed cross-country analysis of the structural transformation process was presented in IMF Country Report No. 15/183, “Raising Productivity Growth in Poland: The Role of Structural Transformation” (July 2015).2 In this analysis, cross-regional labor reallocation is measured with the ratio of employment in each region to total employment, so it does not have to imply a physical migration of employees.The following spillover variables are considered:
International spillovers (foreign direct investment (FDI)). Along with investment, foreign companies facilitate technology transfer and train local employees. Additionally, they stimulate demand for locally supplied goods and subcomponents and foster integration with global supply chains. The share of companies with foreign capital is used as a proxy for FDI in a given region, owing to unavailability of data on actual investment flows by region.
Regional spillovers (labor productivity of neighboring regions). An analysis of spillovers between regions controls for the spatial dimension of data. More productive regions may support growth in neighboring ones by increasing demand for their goods, but also through investment spillovers. The opposite effect may be observed if more productive enterprises cluster in productive regions, and when more skilled workers are drained from less productive regions.
10. Choice of model. The model is estimated using a system GMM estimator6 applied to annual data between 2002 and 2012. It includes time dummies to control for common shocks, and individual effects are removed in this identification strategy. Annual changes in productivity growth are regressed on lagged real productivity level, lagged levels of explanatory variables, and contemporary shifts of employment, with all available lags used as GMM instruments (Appendix II).
11. Empirical evidence points to several factors as being significant determinants of regional labor productivity growth. Regression results are summarized in Appendix I. The determinants of labor productivity growth are discussed below in the order of their significance in the model:
Foreign investment is a significant positive determinant of labor productivity growth. Regression results confirm the intuition that higher presence of foreign investors has facilitated productivity growth. In the base model, a ten percent higher share of enterprises with foreign capital results in a 1.8 percentage point increase in labor productivity growth.
Structural transformation is a significant determinant of labor productivity growth. Results suggest that a 1 percent decline of the share of agricultural employment in total employment adds about 0.5 percentage points to productivity growth. This is due to the structural transformation process discussed earlier, in other words labor reallocation from agriculture to higher-productivity sectors. The findings suggest that labor reallocation to industry results in higher overall productivity growth, but do not show a significant impact of labor reallocation on productivity in the services sector.
Level of educational attainment matters. As expected, educational attainment supports productivity growth. Decreasing the share of population with below secondary education by 10 percent boosts productivity growth by 5 percentage points.
Tighter labor market conditions support productivity enhancements. Availability of idle workers has a negative impact on labor productivity by 0.4 percentage point per 1 percent higher unemployment rate.
Expenditures of local government have an ambiguous impact on labor productivity growth. Investment expenditures of local governments are an insignificant determinant of labor productivity growth. Other expenditures have a positive impact of 0.2–0.5 percentage point faster productivity growth per a 1 percentage point increase in the expenditure to gross value added ratio. All in all, regression results suggest that local government outlays do not matter much for regional productivity dynamics.
Positive spillovers from proximity to regions with high productivity levels are almost absent. The estimated impact of neighboring more productive regions on productivity growth is limited.
Patent applications are not a significant determinant of labor productivity improvements. The number of patents applications is not statistically significant, although this maybe an imperfect measure of innovation. This is because of likely limited use of European patents (EPO) in the analyzed period.
12. An analysis of labor productivity growth by sector shows that structural transformation is particularly important in agriculture, while foreign direct investment is important in industry and services. Further, an analysis by urban and rural clusters7 shows that foreign direct investment and the level of education matter in each cluster. Higher levels of FDI and higher levels of education are correlated with higher labor productivity growth.
D. Conclusions and Policy Recommendations
13. Despite strong economic performance over the last two decades, there are significant and enduring income disparities between western and eastern regions of Poland. These disparities are strongly correlated with labor productivity differences. While labor productivity growth in poorer eastern regions has been driven significantly by structural transformation, in wealthier western regions it has been driven by higher investment and integration with the German supply chain. Education and labor market conditions had a significant impact on labor productivity growth across regions. Similar growth rates in labor productivity across regions have prevented eastern regions from catching up to western regions.
14. The analysis of regional productivity determinants points to policies that could be conducive to regional productivity convergence.
Support structural transformation and boost productivity in agriculture: Significant room remains to boost labor productivity growth within poorer regions by supporting the reallocation of labor from low-productivity agriculture to higher-productivity industry and service sectors (Figure 3). To unleash the potential for such structural transformation, a review of incentives pertaining to employment in agriculture would be appropriate to identify mechanisms that may encourage people to stay in low productivity farms. In particular, the highly subsidized pension scheme for farmers could be reformed to gradually align it with the regular system to discourage inefficient farming motivated by pension arbitrage. The more productive farms in western regions tend to be larger, so there may be merit in promoting consolidation of agricultural production also in the poorer eastern regions to exploit the economies of scale. In this regard, moving from the current agricultural taxation based on farm size and quality of land to an income-based tax would reduce disincentives to scale up farms and help define the base for social security contributions. To facilitate structural transformation, such reforms should be accompanied with measures to address skill mismatches and bottlenecks in labor mobility, as described below.
Figure 3. Poland: Structural Transformation in Agriculture
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Figure 3. Poland: Structural Transformation in Agriculture
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Figure 3. Poland: Structural Transformation in Agriculture
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Encourage labor mobility and reduce structural unemployment: Decomposition of productivity growth shows that the contribution from reallocating labor across regions has been relatively minor. This suggests bottlenecks in labor market mobility that could be addressed with proper policies, for example, by improving the functioning of the housing rental market. Currently, the rental housing market in Poland is shallow, discouraging labor relocation. Econometric analysis also suggests that high structural unemployment negatively affects regional productivity growth. While a declining working age population should generally reduce the unemployment rate, addressing high structural unemployment in less productive regions would require greater investment in active labor market policies to improve job searching efficiency across regions, upgrade skills, and reduce skill mismatches.
Attract investments to less productive regions: Empirical findings show that higher FDI is associated with faster regional productivity growth. FDI is more prevalent in wealthier regions, and this pattern needs to change to support regional productivity catch-up. Some factors important for investors could be altered by government policies to support such a change.8 Specifically, strengthening transportation networks in poorer regions would help, and better targeting of EU funds could support this process. Furthermore, investor surveys suggest that access to skilled labor is important for location of projects. In this context, investing in education and tailoring it to local development needs is important; aligning vocational curricula closely to the needs of industry would facilitate the absorption of new production methods and technologies. While local governments’ role in boosting productivity appears less statistically significant, it does not imply that quality of local administration is irrelevant. For example, data suggest a positive correlation between regional productivity and the efficiency of local tax administration.
Tax offices efficiency and productivity growth
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Eurostat, MoF, and IMF staff calculations.Tax offices efficiency and productivity growth
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Eurostat, MoF, and IMF staff calculations.Tax offices efficiency and productivity growth
Citation: IMF Staff Country Reports 2016, 211; 10.5089/9781475525632.002.A001
Sources: Eurostat, MoF, and IMF staff calculations.
References
Arellano, M., and O. Bover, 1995, “Another look at the instrumental variables estimation of error-components models” Journal of Econometrics 68(1), pp. 29–51.
Arellano, M. and S, Bond, 1991, “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations” The Review of Economic Studies 58(2), pp. 277–297.
Baltagi, B. and N. Pinnoi, 1995, “Public Capital Stock and State Productivity Growth: Further Evidence from an Error Components Model” Empirical Economics 20, pp. 351–359.
Barro, R. and X. Sala-i-Martin, 1991, “Convergence across States and Regions”, Brookings Papers of Economic Activity 1, pp. 107–182.
Baumol, W., 1986, “Productivity Growth, Convergence and Welfare: What the Long-Run Data Show” American Economic Review 76(5), pp. 1072–1085.
Belogrey, N., R. Lecat and T. Maury, 2004, “Determinants of Productivity per Employee: An Empirical Estimation Using Panel Data”, Working Paper (Banque de France).
Bernard, A. and C. Jones, 1996, “Productivity across Industries and Countries: Time Series Theory and Evidence” The Review of Economics and Statistics 78(1), pp. 135–146.
Blundell, R. and S. Bond, 1998, “Initial conditions and moment restrictions in dynamic panel data models,” Journal of Econometrics 87(1), pp. 115–143.
Borsi, M. and N. Metiu, 2013, “The Evolution of Economic Convergence in the European Union”, Deutsche Bundesbank Discussion Paper No. 28/2013 (Deutsche Bundesbank).
Choudhry, M., 2009, “Determinants of Labor Productivity: An Empirical Investigation of Productivity Divergence”, Working Paper (University of Groningen).
International Monetary Fund, 2015, “Raising Productivity Growth in Poland: The Role of Structural Transformation” IMF Country Report No. 15/183.
Polish Agency for Information and Foreign Investments (PAIIZ), HSBC and Grant Thornton, 2015, “Zagraniczny biznes zakorzenil sie w Polsce”.
Sonderman, D., 2012, “Productivity in the Euro Area: Any Evidence of Convergence?” European Central Bank Working Paper No. 1431 (European Central Bank).
Van Der Horst, H. and L. Bettendorf, 2009, “Does Employment Affect Productivity?” CPB Discussion Paper No. 119 (CPB Netherlands).
Van Pottlesberghe, B. and F. Lichtenberg, 2001, “Does Foreign Direct Investment Transfer Technology across Borders?” The Review of Economics and Statistics 83(3), pp. 490–497.
Winiarczyk-Raniak, A. and P. Razniak, 2011, “Regional Differences in the Standard of Living in Poland (Based on Selected Indices)”, Procedia Social and Behavioral Sciences 19, pp. 31–36.
Appendix I. Regression Results
| Dependent variable: Real labor productivity growth | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| VARIABLES | (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) |
| Real Productivity (t-1) | −0.258*** | −0.304*** | −0.353*** | −0.320* | −0.484*** | −0.299*** | −0.461*** | −0.390** | −0.405** |
| (0.003) | (0.001) | (0.001) | (0.068) | (0.000) | (0.002) | (0.000) | (0.011) | (0.011) | |
| Change in share of employment in agriculture (t) | −0.587*** | −0.492*** | −0.455** | −0.517*** | −0.427** | −0.490*** | −0.538*** | ||
| (0.001) | (0.007) | (0.0358) | (0.010) | (0.035) | (0.007) | (0.002) | |||
| Change in share of employment in industry (t) | 0.508*** | ||||||||
| (0.000) | |||||||||
| Change in share of employment in services (t) | −0.187 | ||||||||
| (0.373) | |||||||||
| Share of enterprises with foreign capital (t-1) | 0.184* | 0.2047 | 0.248** | 0.263 | 0.479*** | 0.387** | 0.423** | ||
| (0.075) | (0.042) | (0.025) | (0.116) | (0.007) | (0.027) | (0.031) | |||
| Share of population below secondary education (t-1) | −0.439*** | −0.495*** | −0.553*** | −0.619*** | −1.055*** | ||||
| (0.005) | (0.000) | (0.004) | (0.002) | (0.000) | |||||
| Patents applications per million employed (t-1) | 0.00482 | ||||||||
| (0.830) | |||||||||
| Unemployment rate (t-1) | −0.396*** | −0.320* | −0.293** | ||||||
| (0.004) | (0.096) | (0.016) | |||||||
| Local govt investment share of expenditures (t-1) | −0.0507 | ||||||||
| (0.465) | |||||||||
| Local govt non-investment expenditures to GVA (t-1) | 0.204 | 0.503* | |||||||
| (0.490) | (0.057) | ||||||||
| Neighbors GVA per capita gap (t-1) | 0.0179 | −0.00700 | |||||||
| (0.505) | (0.769) | ||||||||
| Time dummies | YES | YES | YES | YES | YES | YES | YES | YES | YES |
| No. of Observations | 630 | 630 | 630 | 693 | 630 | 630 | 630 | 630 | 630 |
| No. of Regions | 63 | 63 | 63 | 63 | 63 | 63 | 63 | 63 | 63 |
| No. of Instruments | 02 | 70 | 70 | 66 | 70 | 72 | 71 | 73 | 85 |
| AR1 | (0.000) | (0.000) | (0.000) | (0.000) | (0.000) | (0.000) | (0.000) | (0.000) | (0.000) |
| AR2 | (0.156) | (0.038) | (0.012) | (0.143) | (0.100) | (0.119) | (0.082) | (0.114) | (0.100) |
| Hansen test | (0.173) | (0.358) | (0.240) | (0.156) | (0.219) | (0.141) | (0.278) | (0.629) | (0.219) |
| Dependent variable: Real labor productivity growth | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| VARIABLES | (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) |
| Real Productivity (t-1) | −0.258*** | −0.304*** | −0.353*** | −0.320* | −0.484*** | −0.299*** | −0.461*** | −0.390** | −0.405** |
| (0.003) | (0.001) | (0.001) | (0.068) | (0.000) | (0.002) | (0.000) | (0.011) | (0.011) | |
| Change in share of employment in agriculture (t) | −0.587*** | −0.492*** | −0.455** | −0.517*** | −0.427** | −0.490*** | −0.538*** | ||
| (0.001) | (0.007) | (0.0358) | (0.010) | (0.035) | (0.007) | (0.002) | |||
| Change in share of employment in industry (t) | 0.508*** | ||||||||
| (0.000) | |||||||||
| Change in share of employment in services (t) | −0.187 | ||||||||
| (0.373) | |||||||||
| Share of enterprises with foreign capital (t-1) | 0.184* | 0.2047 | 0.248** | 0.263 | 0.479*** | 0.387** | 0.423** | ||
| (0.075) | (0.042) | (0.025) | (0.116) | (0.007) | (0.027) | (0.031) | |||
| Share of population below secondary education (t-1) | −0.439*** | −0.495*** | −0.553*** | −0.619*** | −1.055*** | ||||
| (0.005) | (0.000) | (0.004) | (0.002) | (0.000) | |||||
| Patents applications per million employed (t-1) | 0.00482 | ||||||||
| (0.830) | |||||||||
| Unemployment rate (t-1) | −0.396*** | −0.320* | −0.293** | ||||||
| (0.004) | (0.096) | (0.016) | |||||||
| Local govt investment share of expenditures (t-1) | −0.0507 | ||||||||
| (0.465) | |||||||||
| Local govt non-investment expenditures to GVA (t-1) | 0.204 | 0.503* | |||||||
| (0.490) | (0.057) | ||||||||
| Neighbors GVA per capita gap (t-1) | 0.0179 | −0.00700 | |||||||
| (0.505) | (0.769) | ||||||||
| Time dummies | YES | YES | YES | YES | YES | YES | YES | YES | YES |
| No. of Observations | 630 | 630 | 630 | 693 | 630 | 630 | 630 | 630 | 630 |
| No. of Regions | 63 | 63 | 63 | 63 | 63 | 63 | 63 | 63 | 63 |
| No. of Instruments | 02 | 70 | 70 | 66 | 70 | 72 | 71 | 73 | 85 |
| AR1 | (0.000) | (0.000) | (0.000) | (0.000) | (0.000) | (0.000) | (0.000) | (0.000) | (0.000) |
| AR2 | (0.156) | (0.038) | (0.012) | (0.143) | (0.100) | (0.119) | (0.082) | (0.114) | (0.100) |
| Hansen test | (0.173) | (0.358) | (0.240) | (0.156) | (0.219) | (0.141) | (0.278) | (0.629) | (0.219) |
Appendix II. Econometric Model
The identification strategy is the Blundell and Bond system generalized method of moments estimator (see Arellano and Bover (1995) and Blundel and Bond (1998)). This model is suitable for dynamic panel data where the number of individuals is large relative to time periods, where independent variables may not be strictly exogenous and the idiosyncratic disturbances may have individual-specific patterns of heteroskedasticity and serial correlation. In addition, system GMM produces unbiased estimates even in case of highly persistent data, such as structural variables. Two-step procedure was applied, due to its higher asymptotic efficiency, and standard errors downward bias was corrected by Windmeijer’s adjustment.
Annual changes in productivity growth are regressed against lagged real productivity level and lagged levels of explanatory variables and contemporary shifts of employment, with all available lags used as GMM instruments. Time dummies are included to control for common shocks. Individual effects are removed in transformation of the data within the GMM estimation process.
The bottom panel of Appendix I reports results of standard specification tests. The Arellano-Bond test for second order autocorrelation1 confirms that there is no significant autocorrelation among the instruments. Hansen tests for over-identification restrictions do not reject the null, suggesting that instruments are valid.
The unit of observation is a region-year. Data for 63 regions are included over the period 2002 to 2012. The dependent variable is real labor productivity growth, measured as gross value added per employed (including self-employed), and deflated by the GDP deflator. Data are obtained from Eurostat and the Polish Central Statistical Office.
Independent variables include the following structural variables: change in share of employment in agriculture, industry and services, share of the population with education attainment below secondary level, registered unemployment rate, patent applications per million of employed persons, investment share of expenditures and non-investment expenditures of local governments as percent of gross value added.
To measure spillovers, share of enterprises with foreign capital and neighbors’ value added gap are included. Neighbors’ value added gap measures the difference between average value added per capita in regions neighboring a given region and the given region’s value added per capita, as a percent of the latter.
In addition to the base model, a number of robustness tests were performed. The system GMM model was applied to productivity dynamics in each of three main sectors of the economy, to data transformed into 2-year non-overlapping periods, and cross-section and random effects models were run on the same variables, confirming the results. Additionally, total factor productivity was calculated and used as a dependent variable, within a random effects specification. Results were broadly consistent.
Appendix III. Decomposition of Growth by Sector and Region
The decomposition of productivity growth by sector follows McMillan et al. (2014). The change in aggregate the labor productivity level (Pt) is decomposed as follows:
where i is the sector, L is the number of employed people, and t the period indicator. The second term on the right-hand side quantifies the degree of structural transformation. The change in aggregate productivity is decomposed into a “within” effect (first term in the equation) and a “labor reallocation” effect (second term in the equation). The within effect measures within-sector changes in productivity, while the labor reallocation term measures structural transformation (reallocation of labor across sectors). The within-effect is positive when the weighted change in labor productivity levels in sectors is positive. The structural transformation effect is positive when labor moves from less productive to more productive sectors. Decomposition is based on the division of the economy into the following sectors: agriculture, industry (construction and manufacturing), and services (simple and advanced services).
The decomposition of productivity growth by region follows a similar approach, where i denotes a region instead of a sector. The “within-effect” therefore measures within-region changes in productivity, whereas the “labor reallocation” effect measures cross-regional changes in labor. The labor-reallocation effect is positive if labor is reallocated from less productive to more productive regions.
Prepared by Krzysztof Krogulski, Robert Sierhej, and Aaron Thegeya.
Evidence of overall convergence is inconclusive: panel unit root tests do not show convergence, and there is no evidence of sigma convergence. Additionally, evidence does not show convergence between eastern and western regions. However, there is some evidence of within-region beta convergence, with eastern and western regions converging to different steady states.
Labor productivity is calculated as gross value added per person employed (including self-employed). Labor productivity is a more meaningful measure of regional economic performance as GDP per capita is based on population statistics that may not reflect the actual place of residence and work.
See for example Polish Information and Foreign Investments Agency Survey (PAIIZ, 2015).
Data on central government investments are not available at the regional level.
Regions were grouped into rural, urban and industrialized clusters based on dissimilarities in employment structure.
Factors important to foreign investors are inter alia discussed in Foreign Investments Agency Survey (PAIIZ, 2015).
As expected, we observe first degree correlation in residuals, as system GMM uses differentiated data as instruments. Results of a second degree autocorrelation test yield no evidence of significant autocorrelation among the set of instruments.
