The study is based on a representative sample of 7,783 large and medium‑sized enterprises that took part in a survey conducted by Poland’s Central Statistical Office (GUS)—(PNT-02 questionnaire; Polish version of the Community Innova‑ tion Survey)—over the period of 2008–2010. The enterprises are from NACE sections B‑E.14
A chi‑square method with column proportions and the Bonferroni correction was applied to statistically verify significant differences between active and non-active innovators. Non-active innovators (firms that did not introduce product or process innovation in 2008–2010) constitute the majority of the research sample (N=4,988). The remainder are active innovators (N=2795), firms that introduced process (77.6 %), product (73.5 %), organizational (48.3 %) and marketing innovation (39.6 %). The analyzed sample mainly consisted of medium‑sized enterprises (67.4 %), from medium‑
13 The abbreviations used in the model stand for the following: InnoFundEU – financial support from EU; ExtR&D – acquisition of external R&D; AcqMachEq – acquisition of advanced machinery and equipment; TrainPers – training of personnel for innovative activities; AbsCap – absorptive capacity;
InstCoop– institutional cooperation. Details of variable operationalization are given in Table 9.
14 The selection of units for the survey was performed using the Polish Classification of Activi‑ ties (PKD) 2007, consistent with the Statistical Classification of Economic Activities in the European Union (Nomenclature statistique des Activités économiques dans la Communauté Européenne) (NACE Rev. 2). In 2011, the study on innovation in both industry (Sections B to E) and the service sector (Sections H to M) was conducted on the entire group of entities. For details, see: Innovation Activities of Enterprises in
technology industries (55.7 %) (Eurostat classification, 2008), for which the local market is the most important sales destination (48.6 %).
Table 8 Sample characteristics Sample characteristics Sample in the model N=652 Non‑active innovators N=4,988 Active innovators N=2,795 Whole sample N=7,783 N % N % N % N %
Introduction of product innovation 524 80.4 0 0a 2055 73.5b 2,055 26.4 Introduction of process innovation 480 82.8 0 0a 2169 77.6b 2,169 27.9 Introduction of organizational innovation 414 63.5 458 9.2a 1349 48.3b 1,807 23.2 Introduction of marketing innovation 342 52.5 402 8.1a 1107 39.6b 1,509 19.4 Firm size Medium 331 50.8 4,356 87.3a 1885 67.4b 6,241 80.2 Large 321 49.2 632 12.7a 910 32.6b 1,542 19.8 Technology level
Not classified 68 10.4 655 13.1a 272 9.7b 927 11.9 Low tech 95 14.6 2,232 44.7a 843 30.2b 3,075 39.5 Medium tech 440 67.5 2,026 40.6a 1,558 55.7b 3,584 46 High tech 49 7.5 75 1.5a 122 4.4b 197 2.5 Dominant market
Local 588 90.2 1,667 33.4a 661 23.6b 2,328 29.9 Domestic 581 89.1 1,981 39.7a 1,359 48.6b 3,340 42.9 EU 524 80.4 1,165 23.4a 654 23.4a 1,819 23.4 Other markets 412 63.2 175 3.5a 121 4.3a 296 3.8 a Each subscript letter (a, b) denotes a subset of categories whose column proportions (Bonferroni method)
differ significantly from each other at the.05 level.
Source: Own calculation in SPSS 21 based on data from PNT‑02 questionnaire, Sprawozdanie o innowacjach
w przemyśle za lata 2008–2010, www.stat.gov.pl/formularze
Due to the PNT‑02/CIS questionnaire construction, where most questions refer to innovative enterprises, we will assume, like other researchers (Veugelers, Cassi‑ man, 2004; Mothe et al., 2010), as a filter variable indication of whether the company introduced new or significantly improved products or processes in 2008–2010.
In addition, we assume that only companies that received public support for innovative activity in the researched period will be analyzed. Based on this we extract 652 companies. Details on the operationalization of all the variables are presented in Table 9.
Table 9
Variable operationalization
Variable Description and construction of variables
InnoActComp Filter variable – “Innovation activity” and “Public support”
InnoActCompPr “1” if the firm introduced product innovation; “0” otherwise and/or
InnoActCompProc “1” if the firm introduced product innovation; “0” otherwise
InnoFund “1” if the firm received public financial support from local agencies, government agencies or EU
InnoFundEU Variable – “Financial support from EU”
Calcuated if the firm received public financial support for innovation activity from EU for personnel training; support of international cooperation; support of domestic, regional, cluster cooperation; support of exporting; specialized consulting; support for investments; support for cooperation with institutional partners; support for R&D activity; other programs.
InnoPerf Dependent variable – “Innovation performance”
Log of fraction (from 0 to 100) of turnover from innovative products introduced in 2008–2010 in total turnover in 2010.
InnoExp Variables – “Expenditures on innovation activities”
ExtR&D
Calcuated if the firm declared acquisition of external R&D and/or acquisition of external knowledge (purchase or licensing of patents and non‑patented inventions, know‑how, and other types of knowledge from other enterprises or organizations for the development of new or significantly improved products and processes).
AcqMachEq “1” if the firm declared acquisition of advanced machinery or equipment (including computer hardware) or software to produce new or significantly improved products and processes; “0” otherwise.
AbsCap Variable – “Absorptive capacity”
“1” if the firm performed R&D continuously (had permanent R&D staff in‑house) from 2008 to 2010; “0” otherwise.
TrainPers Variable – “Training for innovative activities”
“1” if the firm conducted internal or external training for its personnel, specifically for the development and/or introduction of new or significantly improved products and processes; “0” otherwise.
InstCoop Variable – “Cooperation with institutional partners”
Calcuated if the firm declares cooperation with the Polish Academy of Sciences; domestic research institutes; domestic universities; foreign research institutes; foreign universities.
Source: Own compilation based on PNT‑02 questionnaire for 2008–2010, www.stat.gov.pl/formularze The structural equation modeling (SEM) method—specifically a technique known as path analysis, designed to examine the structure and strength of linear relationships between at least one independent variable and one or more dependent variables—will be used to assess the relationships between the variables (Bedyńska, Książek, 2012). The aim of SEM is to find a model that describes reality in the best way (Perek-Białas,
Pleśniak, 2013). In order to verify the hierarchy of variables, an analysis was conducted of critical values between parameters.
Since reasoning based only on data from a single sample may result in an over‑ or under‑estimation of the parameters of the population, the analysis of the distribution of the estimation errors was made with multiple sampling with replacement from the sample (non‑parametric bootstrap method) (Hayes, 2009; Efron,1979). The models applied the Bollen and Stine (1992) correction for the p level to test the null hypoth‑ esis of model fit.