DEM_2015_157to165 © 2015 Nicolaus Copernicus University. All rights reserved. http://www.dem.umk.pl/dem D Y N A M I C E C O N O M E T R I C M O D E L S DOI: http://dx.doi.org/10.12775/DEM.2015.008 Vol. 15 (2015) 157−165 Submitted October 24, 2015 ISSN (online) 2450-7067 Accepted December 15, 2015 ISSN (print) 1234-3862 Tomasz Groszkowski, Tomasz Stryjewski* The Econometric Analysis of Entrepreneurship Determinants in Polish Voivodeships in the Years 2004–2013 A b s t r a c t. This article presents research results and describes and clarifies differences in a level of the entrepreneurship index observed between Polish voivodeships in the period from 2004 to 2013. The expected results were confirmed with the fixed effect (FE)/random effect (RE) panel data model. The analysis showed that the level of entrepreneurship varies depending on a voivodeship, and that this differentiation is structural and permanent. The applied model also confirmed the expected determinants of entrepreneurship. K e y w o r d s: entrepreneurship, development factors, panel data model, FE/RE model. J E L Classification: C23, C26, L26, R12. Introduction Entrepreneurship is undoubtedly one of the fundamental economic de- velopment factors, through its influence on development of new products, new markets, creating jobs, and raising the social welfare in general. It is noticeable that some areas are growing fast through establishment of new companies, when others remain far behind. The concept of entrepreneurship is a complex issue, and it is permanently inscribed into different branches of science. A wide range of studies shown that a large and continuously in- creasing number of classifications and descriptions of entrepreneurship re- * Corresponding author: Tomasz Stryjewski, The University of Computer Science and Economics in Olsztyn, e-mail: tomasz.stryjewski@erbud.pl; Tomasz Groszkowski, The Uni- versity of Computer Science and Economics in Olsztyn. Tomasz Groszkowski,Tomasz Stryjewski DYNAMIC ECONOMETRIC MODELS 15 (2015) 157–165 158 sults in a lack of a uniform theory of entrepreneurship and associated measures, implying possible difficulties in their assessment (Szarecki, 2008, p. 181). Entrepreneurship is multidimensional and consists of many elements that should be considered when determining the characteristics of entrepre- neurship (Kalkan, Kaygusuz, 2012). A management sphere pays particular attention to this aspect, mainly for functional reasons, describing it as the process of organizing and running a business in conditions of risks associated with those activities (Gryffin, 1997, p. 730–731). Entrepreneurship can therefore be considered as a process creating or identifying opportunities, which later are frequently used regardless of re- sources possessed at a given time. Entrepreneurship understood this way is a stimulating factor for a creative entrepreneur who finds energy to establish and build a company or organization. Therefore, the entrepreneur is more than just a passive observer of the situation in which he is positioned (Tim- mons, 1990, p. 5). Entrepreneurship can therefore be described as an orga- nized process consisting of successive phases, occurring in specific condi- tions and aiming at using innovative ideas to achieve certain benefits, while considering risks involved in this process (Adamczyk, 1995, p. 9–10). Entrepreneurship is one of the basic factors of economic, social and cul- tural development. It is associated with activities of different types of busi- nesses, from microenterprises, throughout the whole SME sector, up to large domestic and international enterprises. Certainly, companies play an im- portant role in the Polish and in the European economies. They create a space where human skills and entrepreneurial attitudes are revealed and concentrated. Furthermore, they are a source of innovations and employ- ment. The experience of well-developed countries shows that entrepreneur- ship plays a significant role in the economy, stimulating economic growth, influencing employment, and providing goods to the market. Therefore, companies are a very important factor for economic growth of a relevant area. Entrepreneurship and businesses are also stimulating factors for growth, and the number of enterprises is frequently considered to be an indi- cator of economic development. 1. Entrepreneurship Factors Entrepreneurial individuals, stimulated into action in appropriate condi- tions, establishing businesses, creating new jobs and additional sources of income for themselves and the local population, can be found in every com- munity, also at a local level. The Econometric Analysis of Entrepreneurship Determinants… DYNAMIC ECONOMETRIC MODELS 15 (2015) 157–165 159 However, an answer to a question about factors influencing entrepre- neurial behavior is not so clear and obvious. In numerous studies on entre- preneurship, selection of factors that affect the level of entrepreneurship is a frequently encountered problem. This article focuses on economic factors, describing the structural and economic determinants of entrepreneurship and quantitative aspects of that issue. Entrepreneurship is determined by the certain socio-economic and politi- cal factors of intensity varying in time and space. Those factors may act in two ways: by bringing positive effects to the economy in form of incentives for business and economic growth stimulators or, on contrary, as means re- straining or even inhibiting creation of new and development of already ex- isting business entities. These factors are classified according to various criteria (Chrapek, 2009, p. 321). Their effects on a recovery of a given area can either be positive or ad- verse, through their potential to create conditions for formation and devel- opment of entrepreneurship. Providing suitable conditions for development of entrepreneurship should therefore be one of the most important priorities for the regional development for authorities at any level (Jezierska-Thole, 2010, p. 129). Processes associated with entrepreneurial behavior occur in specific so- cio-economic circumstances. GDP, an economic situation, revenues or in- vestments are just examples of certain macroeconomic indicators affecting development of enterprises. Considering the entrepreneurship development conditions, it should be noted that every company operates in a certain environment with which it interacts. In entrepreneurial research, one of the common problems is identi- fying factors determining an entrepreneurial activity. A type of analyzed data, constrained a limitations in very narrow definition of entrepreneurship as the number of formal registered companies. In this study, it was decided to apply a division into regions defined as established territorial units of a relatively large surface area and population, and with a specific economic policy prevailing. In Polish economic reality, this region can correspond to a voivodeship. This approach is based on factors including their independently established poli- cies, independent local authorities, and own budget. With specifics of each voivodeship, an endogenous development potential can be developed, influ- encing opportunities and barriers to the development and growth of enter- prises in that region. According to the subject of this article, individual features of the region, having a direct influence on the investment level and the profitability of the Tomasz Groszkowski,Tomasz Stryjewski DYNAMIC ECONOMETRIC MODELS 15 (2015) 157–165 160 business should be emphasized (Godlewska, 2001, p. 14). If the region is understood as an environment consisting of humans, other companies and institutions, having certain features that may or may not be attractive to new businesses, then it is important to communicate conditions prevailing in that region, which will affect the company's position, and to ensure future entre- preneurs are familiar with and have access to that information. 2. Research Methodology and Analysis Results In this paper, the main part of the space-time analysis of entrepreneur- ship in Poland aimed at finding information about factors determining spe- cific indicators, and reasons why these indicators vary in individual voivodeships. A panel approach of the fixed effect/random effect (FE/RE hereafter) type is suitable for such kind of modeling (Greene, 2005). The analysis covered annual data for a period from 2004 to 2013 for 16 Polish voivodeships. 15 factors possibly affecting the entrepreneurship de- velopment in Poland1 were selected, following a detailed substantive analy- sis: Gross Domestic Product per capita (GDP per capita), enterprises invest- ment outlays per capita (inv), retail sales per capita (sprzed), new apartments put in use per thousand inhabitants (mieszk), population density (gest), per- centage of population living in cities (urbanizacja), number of cities in a region (miasta), roads outside built-up areas per 100 km2 (drogi), employ- ment rate (zatr), average monthly disposable income per capita (doch), num- ber of tourist facilities per capita (turyst), number of people with higher education per thousand inhabitants (wykszt), net migration per thousand inhabitants (migr1000), and municipal revenues per capita (doch per capita). Starting with a framework of the congruent modeling concept, an inter- nal structure of specific processes was analyzed during the first stage of the study (Talaga, Zieliński, 1989). A congruent panel data model exhibits the same harmonic structure on both sides of the equation in dynamic terms. The analysis showed the significance of a linear trend in all tested processes and autoregressive relationships over time. However, due to a small sample size in the researched period, only the first-order autoregression was considered. The main problem associated with the use of models of this class is that they belong to a group of static methods. Modeling the dynamic relationship with these methods results in both OLS and GLS coefficients bias (Wooldridge, 2010). 1 Understood as the number of companies per 10000 people of employable age (entrepre- neurship index – wskpz). The Econometric Figure 1. Entrepreneurship index 2004–2013 Unfortunately, dynamics of terms of ACF and PACF functions) ble is not allowed in the FE/RE models priate instrumental variable that the search for an give any meaningful cial" variable correspond An instrumental variable for the entrep ed with a dynamic panel model based on (model hypothesis): itwskpz α= 0 where: ������� – spatial entrepren trend, � ,��,��,� Econometric Analysis of Entrepreneurship Determinants… DYNAMIC ECONOMETRIC MODELS 15 (2015) Entrepreneurship index – time series for voivodeships in the years dynamics of all processes exhibits strong autocorrelation terms of ACF and PACF functions). Since use of a lagged dependent vari in the FE/RE models, it was necessary to select an priate instrumental variable, so no information was lost. It should be noted an instrument within a specified group of variables meaningful results. Therefore, it was necessary to create an corresponding to a character of the real instrument. nstrumental variable for the entrepreneurship indicator was a dynamic panel model based on a structure of internal variables : ittiti twskpzwskpz ηααα ++++ −− 32,21,10 spatial entrepreneurship indicator in period t, � – evaluation parameters, η �� – random spatial comp ) 157–165 161 time series for voivodeships in the years autocorrelation (in dependent varia- select an appro- It should be noted specified group of variables did not an "artifi- reneurship indicator was evaluat- of internal variables (1) – linear random spatial compo- Tomasz Groszkowski,Tomasz Stryjewski DYNAMIC ECONOMETRIC MODELS 15 (2015) 157–165 162 nent in period t. The following model was developed with a two-step estima- tion method2 (2-step method). Table 1. A dynamic panel model for a spatial entrepreneurship indicator (Blundell and Bond estimator) Dependent variable: wskpz Coefficient SD z p-value wskpz(-1) 0.9835 0.0385 25.5764 <0.00001 *** Const. 1.4756 5.0447 0.2925 0.7699 Time 0.4533 0.1128 4.0188 0.00006 *** Sargan over-identification test: Chi-square(43) = 15.7061 Test for AR(1) errors: z = –2.27161 [0.0231] Test for AR(2) errors: z = –0.713604 [0.4755] The fitted values from the above model created an instrumental variable for a dynamic relationship in FE/RE models. Further research focused on identifying determinants for the spatial en- trepreneurship indicator and its spatial differentiation described by a neigh- borhood matrix. For this purpose, a group of potential exogenous variables was determined and a pooled OLS model was estimated. Results of this es- timation are shown in Table 2 (only statistically significant variables). The next step of the analysis was verification which approach is correct: pooled OLS or FE/RE. Results of this verification are shown in Table 3. Table 2. The data panel model estimation using pooled OLS Dependent variable: wskpz Coefficient SD t-ratio p-value Const. 4.6527 3.59303 1.2949 0.1978 urbanizacja −16.5172 4.61591 –3.5783 0.0005 *** turyst(-1) 0.0625 0.0300 2.0804 0.0396 ** doch per capita 0.0090 0.0018 4.9687 <0.00001 *** Time −0.4339 0.1932 -2.2453 0.0266 ** yhat17(-1) 0.9631 0.0297 32.3751 <0.00001 *** Mean dependent var 149.7844 S.D. dependent var 25.9606 R-squared 0.9824 Adjusted R-squared 0.9817 Log-likelihood −339.4414 Akaike criterion 690.8827 Schwarz criterion 707.9949 Hannan-Quinn 697.8355 rho 0.3696 Durbin-Watson 1.0856 Note: That 17 – instrumental variable from table 1 model. 2 See footnote 1. The Econometric Analysis of Entrepreneurship Determinants… DYNAMIC ECONOMETRIC MODELS 15 (2015) 157–165 163 Table 3. Panel models diagnostics3 F test (Fixed effect) the null hypothesis that the pooled OLS model is adequate, in favor of the fixed effects alternative F(15, 107) = 2.6292 p-value = 0.0021 Breusch-Pagan test (Random effect) the null hypothesis that the pooled OLS model is adequate, in favor of the random effects alternative LM = 1.7714 p-value = prob(chi-square(1) > 1.77137) = 0.1832 Hausman test the null hypothesis that the random effects model is consistent, in favor of the fixed effects model H = 14.5289 p-value = prob(chi-square(5) > 14.5289) = 0.0126 Results presented in Table 3 imply that a fixed effects-FE model is an appropriate approach. This means that the spatial differentiation in the entre- preneurship index level is of a structural and permanent character. Therefore, the next step of the study was to estimate and validate the FE model. The model presented in Table 2 became a model hypothesis and the estimated (after verification) FE model is presented in Table 4. Table 4. The FE model (Dependent variable: wskpz) – LSDV estimator Coefficient SD t-ratio p-value Const 31.9812 9.3363 3.4255 0.00086 *** doch per capita 0.0091 0.0017 5.4573 <0.00001 *** yhat17(-1) 0.7004 0.0728 9.6190 <0.00001 *** Mean dependent var 149.7844 S.D. dependent var 25.9606 Sum squared resid 1127.241 S.E. of regression 3.2012 LSDV R-squared 0.9868 Within R-squared 0.7427 LSDV F(17, 110) 484.8430 P-value(F) 4.79e-95 Log-likelihood −320.8560 Akaike criterion 677.7120 Schwarz criterion 729.0486 Hannan-Quinn 698.5703 Rho 0.2742 Durbin-Watson 1.1935 The FE model has better statistical properties than the model estimated by pooled OLS. In particular, significant reduction in the first-order random component autocorrelation was possible in the FE model. However, the study also showed that the analysis of the entrepreneurship spatial differenti- ation using the spatial entrepreneurship indicator is biased and, in conse- 3 See footnote 1. Tomasz Groszkowski,Tomasz Stryjewski DYNAMIC ECONOMETRIC MODELS 15 (2015) 157–165 164 quence, has some disadvantages. Particularly, this indicator is strongly auto- regressive, indicating an autonomous nature of changes in that process. It may also imply that variables used in the analysis were unsuitable. However, the results presented in Table 4 confirm the economic theory about the entrepreneurship development index. The main determinants of entrepreneurship remain the same, being the entrepreneurship indicator from the previous period and the municipal revenues per capita. These two varia- bles positively influence the entrepreneurship rate, although they strengthen the autoregressive character of that process. Conclusions A study on entrepreneurial activities based on their index led to three main conclusions. First, the analysis showed that the problem of entrepreneurship devel- opment in Poland is a structural problem. This means that differences in the levels in individual regions are stable and result from internal conditions, as well as from the regional polarization processes. Therefore, this problem cannot be solved locally, but only as part of a coherent regional policy at the governmental or European level. Second, the entrepreneurship index determinants in the studied period were identified. The main determinant is the municipal revenues per capita. However, the entrepreneurship index is a strongly autonomous variable and, therefore, the autoregressive component is important in the analyzed model. The main causes of the dependent variable formation are the regional eco- nomic situation (autoregressive process) and public revenues. Public reve- nues affect the dependent variable in two ways: as a determinant (indicator) of regional economic situation and as the basis for public expenditures which, in turn, influence small local enterprises. Third, an important part of the study is focusing on a selection of suita- ble explanatory variables. The spatial entrepreneurship indicator is highly autocorrelated, and this causes problems with identification of its determi- nants. In the further research, other measures should be considered for explo- ration and selection, better describing the investigated process. References Adamczyk, W. (1995), Ewolucja form i typów przedsiębiorczości (The evolution of forms and types of entrepreneurship), Zeszyty Naukowe Akademii Ekonomicznej w Poznaniu, 236. The Econometric Analysis of Entrepreneurship Determinants… DYNAMIC ECONOMETRIC MODELS 15 (2015) 157–165 165 Chrapek, G. (2009), Przedsiębiorczość osób fizycznych na obszarach wiejskich Podkarpacia, [w:] Rola przedsiębiorczości w kształtowaniu społeczeństwa informacyjnego (Entre- preneurship individuals in rural areas of Podkarpacie, [in:] the role of business in shaping the information socjety), Seria Przedsiębiorczość – Edukacja, Warszawa- Kraków. Godlewska, H. (2001), Lokalizacja działalności gospodarczej (The location of economic activity), Dom Wydawniczy ELIPSA, Warszawa. Greene, W. H. (2005), Econometric Analysis, Prentice Hall, New York. Gryffin, R. W. (1997), Podstawy zarządzania organizacjami (Fundamentals of organizational management), PWN, Warszawa. Jezierska-Thole, A. (2010), Zmiany poziomu infrastruktury i jej wpływ na rozwój przedsię- biorczości na obszarach wiejskich na przykładzie województwa kujawsko- pomorskiego i pomorskiego (Changes in the level of infrastructure and its impact on development of entrepreneurship in rural areas on the example of Kujawsko- Pomorskie and Pomorskie), Acta Acientum Polonorium, Oeconomia, 9(3). Kalkan, M., Kaygusuz, C. (2012), The Psyhology of Enterpreneurship [in:] Entrepreneurship – Born, Made and Educated, Ed. T. Burger-Helmchen, Rijeka. Szarecki, A. (2008), Przedsiębiorczość jako forma kultury (Entrepreneurship as a form of culture), Problemy Zarządzania, 2. Talaga, L., Zieliński, Z. (1986), Analiza spektralna w modelowaniu ekonometrycznym (Spec- tral analysis in econometric modeling), PWN, Warszawa. Timmons, J. (1990), New Venture Creation, Irvin, Boston. Wooldridge, J. M. (2010), Econometric analysis of cross section and panel data, Massachu- setts Institute of Technology. Ekonometryczna analiza determinant przedsiębiorczości w Polskich województwach w latach 2004–2013 Z a r y s t r e ś c i. W artykule przedstawiono wyniki badań, opisujące i wyjaśniające różnice w poziomie wskaźnika przedsiębiorczości pomiędzy województwami polski w latach 2004– –2013. W celu potwierdzenia zakładanych rezultatów, użyto modelu panelowego z efektami stałymi i losowymi. Analiza wykazała, że wskaźnik przedsiębiorczości jest zróżnicowany pomiędzy województwami, a zróżnicowanie to ma charakter strukturalny i stały. Model po- twierdził również oczekiwane determinanty przedsiębiorczości. S ł o w a k l u c z o w e: przedsiębiorczość, czynniki rozwoju, model panelowy, model FE/RE Pusta strona