Exploration of the Effectiveness of Public Procurement for Innovation: Panel Analysis

Exploration of the effectiveness of public procurement for innovation: Panel analysis of EU countries’ data

Detelj Kristina a, Jagric Timotejb, Markovic-Hribernik Tanjab

aFaculty of Organization and Informatics Varazdin, University of Zagreb, Pavlinska 2, 42000 Varazdin, Croatia

bFaculty of Economics and Business, University of Maribor, Razlagova 14, 2000 Maribor, Slovenia

Abstract:

This research focuses on the impact of public procurement for innovation (PPI) on a country's level of innovativeness. The available literature primarily consists of case studies that identify PPI’s impact on the innovativeness of particular firms. Therefore, this paper developed an econometric model to investigate the impact of PPI on the innovativeness of EU countries. The model tested the impact of four different innovation policy measures (PPI, R&D subsidies, regulations and cooperation). Theresults showed that in different model settings,PPI was positively and significantly related to countries’ innovativeness, whereas the other three measures showedlow significance. These research findings may be important to policy makers when selecting appropriate measures for promoting innovation and thereby alsoenhancing theircountry’s competitiveness.

Key words: public procurement, innovation policy instruments, panel data analysis, EU

1.INTRODUCTION

Due to the current economic and financial crisis, public finances today face reductions in tax revenues which, together with globalization, tax competition and new technologies, force governments to reduce inefficient public spending. This also prompts the creators of economic policies to make the role of the state more focused. In this context, one question becomes increasingly important: which state policy instruments promote innovation better, as innovativeness is generally recognized as an important factor in the competitiveness of an economy(WEF, 2013).

In the analysis of policies which promote innovation, Edler and Georghiou (2007) find that European policies have so far focused almost exclusively on the supply side, meaning that Europeangovernmentspromote innovations by supporting their creators. The other option would be to encourage the demand side by supporting the users of innovations, which opens innovation opportunities in the market and engages innovators in potentially profitable activities. These authors, together with other researchers (Edler et al., 2012, Georghiou et al. 2014; Lember et al. 2015), continue to raise the awareness of the importance of demand-side policies and the need for their inclusion into innovation policies.

Since in the past the orientation towards supply-side innovation policieswas much stronger, this paper also takes into account the demand-side policies in research. At the centre of our research is public procurement, which can serve as a powerful policy measure for governments to promote innovation. Edler and Georghiou (2007) call it public procurement for innovation (PPI). It is based on the notion that the demand can trigger and accelerate the development and diffusion of innovation (Edler et al., 2012).

The paper begins with a brief review of the relevant literature on the topic of PPI and an explanation of the circumstancesthatcaused a rising interest in PPI. Firsly, the paper present the results of the qualitative analysis of PPI cases described in recent research. After that the paper presentsamodel that was created to test the impact of the four innovation policy measures on countries’innovativeness level. The analysis is based on panel data for 28 EU countries in the period from2004 to 2011. After presenting and discussing the results of the analysis,directions for possible future research are proposed in the conclusion.

2.LITERATURE REVIEW

The majority of reviewed empirical research on PPI consists ofqualitative case studies of individual examples in which innovations were the result of public procurement. However, there are very few quantitative studies, which are generally used to verify theoretical models and the response of certain variables in practice.

Unlike regular public procurement,in a more narrow sense,PPI is a process in which some public entity awards a contract to another organization for a product or service that does not yet exist (Edquist et al., 2000). Therefore it is necessary to engage in research and development (R&D) and to create an innovation before delivery. This is why the contracting authority must determine the function of a product or system, rather than the product itself (Georghiou et al., 2003). In a broader sense, PPI can turn governments into leading users of innovations, which diffuses existing innovations and thusimproves their commercialization (Edler and Georghiou, 2007).

Inthe early 1980s, many studies argued that demand was an important incentive for innovation, based on the review of Mowery and Rosenberg (1979). At that time the United States wasconsiderably more technologically advanced than Europe, so Rothwell and Zegveld (1981) recommended that Europe should establish a homogenous common market.In that way,emerging new technologies couldbe promoted with the help of the public procurement system,due to the reduced level of uncertainty for potentially innovative firms, since it reflects true needs and represents a stable demand. After this article, researchers mainlylost interest in PPI until the end of the last century. Then the idea was revived by Edquist et al.(2000), who found that, in practice, in the 1990s, most policies for stimulating innovation were directed towards the supply side (such as subsidies for R&D), whereas the demand side (where PPI belongs) was heavily neglected. As demanding clients are a critical driving force in creatinginnovations, the complex requirements of PPI support the propensity of companies for R&D investment and their propensity to create innovation (Nyiri et al. 2007). After the year 2000, there are increasing examples of interest in the field of PPI, particularly at the EU level (Edler et al., 2005; Aho et al., 2006; EU and OECD, 2011).

The EU itself started and funded most of this research at the beginning of the 21st century (mostly reviewed in Edler and Georghiou, 2007; or Aschhoff and Sofka, 2009). There are also some newer studies,such as DETE 2009; EC DGEI 2011; EU and OECD 2011. This encouraged many national and international research institutions to do the same (e.g., FORA and OECD, 2009; OGC and DBIS, 2009; MEE 2010), as well as caused the number of research articles to rise (e.g., Edler and Georghiou, 2007; Aschhoff and Sofka, 2009; Uyarra and Flanagan, 2010; Kattel and Lember, 2010; Edquist and Zabala-Iturriagagoitia, 2012; Guerzoni and Raiteri, 2012, Zelenbabic 2015). All of these explore different aspects and examine the potential of public procurement to influence the formation of innovations in the economy.

The analysis of the58cases from the above-mentioned literature (see in more detail in Detelj, 2015) shows that one part of those PPI cases includes general policies aimed at creating a supportive environment for PPI (20 cases) and the second part includes innovations resulting from the participation in public procurement (38cases). The cases come primarily from the USA, United Kingdom, Germany, the Netherlands, and the Nordic countries, which does not generally imply that there are no such examples in other countriesbut shows a better awareness of PPI possibilities in these countries. Research on public procurement in Central and Eastern Europe still focuses on some other issues, such asthe problem of corruption in Croatian public procurement (Ateljević and Budak 2010)enhancing the efficiency of public procurement by centralizing procurement activities in Serbia (Jovanovic et al. 2013) or the emergence of innovative public social services in Slovakia, which is induced by non-governmental actors (Merickova et al. 2015). None of this research actually pays attention to PPI.

In the review of the available literature, the paperfound only three quantitative studiesthatexamined PPI. The first is the research on the impact of public procurement and other selected policy measures on innovative outputs, carried out on a sample of German companies (Aschhoff and Sofka, 2009). It showed that PPI and cooperation with others in developing innovations were the two statistically significant policies for generating revenue from innovative goods. Another is the research on the impact of PPI and R&D subsidies on innovation inputs (R&D investment) and innovation outputs (revenue from innovative products and services), conducted on a sample of European companies (Guerzoni and Raiteri, 2012). The impact of PPI on both innovation inputs and outputs is stronger than the impact of R&D subsidies, but these two policies affect innovation best when used together. The third study included public procurement in an economic growth model for the American states and examined how the public demand for high-technology products or services influenced corporate R&D investment, in turn leading to innovation (Slavtchev and Wiederhold, 2011). The results show that the government demand for innovative products and services increases private R&D investments.

The literature review shows an obvious prevalence of qualitative studies over quantitative ones. There is also one other gap:no quantitative studies exist for European countries on the macro level (i.e., the country level). This was the rationale behindour research—to add further empirical evidence about the impact of PPI on EU countries’ innovativeness.

3.RESEARCH MODEL

Since quantitative studies on PPI are very scarce, the paper could not find an adequate macro-level model to testthe assumptions made. Based on the theoretical background of Aschhoff and Sofka's (2009) micro-level study (i.e., the level of enterprises in Germany), the paperbuilt aneconometric model to testhow PPI, combined with the other three policy measures, affects a country’s innovativenessat the macro level of 28 EU countries. Thus, the paperbuilt an econometric model that includedPPI, R&D subsidies, cooperation with others for innovation development, and regulation as inputs, and a country’s innovativeness measure as an output. This was empirically tested in the research.

The usual measure of innovative activity has always been patent application, but that is not a very good measure for many industries (Arundel and Kabla, 1998). For this reason, Fagerberg and Srholec (2008) argue that the measurement of innovation based on patents is often misleading because it measures only the global inventions, while the "small" and incremental innovations, representing the majority of modern innovation activity, are not covered by such measurements. Similarly, Rodriguez and Montalvo (2007) emphasize that the focus should also be extended from research, which is seen as a primary source of innovation, to the development anddiffusion of innovation. Due to the interconnectedness of modern technologies with all the spheres of modern life, innovation is now usually incremental and not radical (Markard et al., 2012). This is why it was decided that the dependent variable in the model would be an indicator of business sophistication and innovation (BS&I) from competitiveness research by the World Economic Forum (WEF, 2013), representing the outputs in the presented econometric model. The BS&I index includes a business sophistication pillar, which covers the knowledge, skills and working conditions embedded in the organizations; and the technological innovation pillar, attached to the traditional view of innovation as new products, services and processes (Sala-i-Martin et al., 2011). The index ranges from 1 to 7, with 7 representing the best result.

The basic regression model is as follows:

(1)

The explanatory variables in our focus are the four policy measures through which the state can encourage the formation of innovation. These are:

• COOP - the proportion of firms that cooperate with others in the process of creating innovation
• RQ - regulatory quality
• GBERD - share of Gross Business Expenditures for R&D financed by the public sector (subsidies)
• PPI - a measure of public procurement for innovation

3.1.Choice of indicators for explanatory variables

Cooperation with others in the creation of innovation (COOP) was represented by Eurostat data (n.d.), based on the Community Innovation Survey (CIS). COOP represents the share of enterprises in a country that create their innovations in cooperation with others. Participants in this cooperation may be manifold (customers, suppliers, competitors, research institutes, universities,etc.). The impact of regulatory conditions on businesses is measured by an indicator fromWorldwide Governance Indicators (WGI), examining the quality of governance. The Regulatory Quality (RQ) "captures the perception of the ability of the government to formulate and implement sound policies and regulations that permit and promote private sector development" (WGI project, 2012). It is a proxy measure for the characteristics of the regulatory environment inEU countries. This indicator reflects changes in business conditions affected by changes in the regulatory framework. Public subsidies in R&D are calculated based on Eurostat’s science and technology indicators. The first measures the total Business Enterprise R&B Expenditure(BERD)and the second shows the amount of euros spent on BERD coming from public funds. Their quotient is our indicator, which shows the share of Gross Business R&D Expenditures financed by the public sector (GBERD). For most of the countries the data was available for the years 2004-2011.

Determining the most appropriate indicator for monitoring PPI proved to be the biggest problem. The EU prescribes regulatory thresholds above which the public entities are obliged to implement a mandatory procurement process (EC, 2013). But, there is no coordination in collecting the data about performed procurement above the thresholds, let alone below them (Kapff, 2013). In the published literature to date,there are no relevant indicators, except for the estimate of the total amount of public procurement in the EU and the share of public procurement in GDP, published by the European Commission (EC, 2009, 2012). The data for the years 2004-2011 is available for the 25 EU Member States,but has only been availablefor Bulgaria and Romania since the year 2007, while for Croatia these reports contain no data. The data for Croatia was thus calculated based on the figures listed in public procurement annual reports of the Croatian Ministry of Economy (MINGORP, 2013) divided by GDP(Worldbank n.d.). Croatian datais also onlyavailable from the year 2007 onwards.

However, the data on total public procurement (PP) does not distinguish between regular and innovative procurement, which would be more appropriate. Therefore, the papercreates acombined variable which consists of thePP totalmultiplied by a dummy indicator reflecting a government’s propensity to use PP for acquiring innovative goods. The dummy was derived from the results of theExecutive Opinion Survey (EOS), where the respondents had to answer whether their governments' decisions in public procurement promoted innovation (WEF 2013). Responses ranged from 1 (no, not at all) to 7 (yes, extremely efficiently). Since this indicator is also a component of the BS&I index, which is our dependent variable, authors could not include it directly into our model as a numeric variable. Therefore, authors used it to create the dummy variable (DPPI) based on the average value of the indicator for the country in a specific year. The countries with results below a certain threshold were given the value of the dummy variable DPPI=0 and the ones with results above the threshold value were given a DPPI=1. This way, the authors created two variables to distinguish between more and less innovative public procurement: PPI_D1 where the dummy variable has taken a value ​​of 1 where the results from EOS were above 4.0, and PPI_D2 has taken a value of 1 where the results from EOS were above 4.5.

The econometric model therefore includes two variables instead of one formonitoringthe PPI:

(2)

By including (2) in our basic model (1), it becomes the following:

(3)

The (un)availability of data for Bulgaria, Romania and Croatia for the whole period (2005-2011) resulted in an unbalanced panel.

In addition to these variables the outputs can also be affected by other factors, which the paper discuss in the extended model as control variables. The authors examine how the BS&I index reflects the impact of the following: GDP per capita (GDP_pc), GDP growth rate (GDP_gr), gross expenditures on R&D as a percentage of GDP (GERD), government financing share in GERD (G_GERD), share of employment in high-technology and knowledge-intensive activities in the economy (TRD_emp and RES_emp), export of high-technology products (HTP_xp) and the export of knowledge-intensive services (KIS_xp ). The goal was to find out which among them significantly affected the dependent variable. Moreover, the literature review suggests that the effects of certain innovation policy measures (e.g. public procurement or changes in legislation) could only be seen after a longer period of time (medium to long term) and not immediately after their use (Edler et al. 2012). Considering this, our econometric model also tested the effects of the explanatory variables with a one-year time lag. Unfortunately, short time-series data availability does not allow us to test our models for longer time lags.

3.2.Methodology and data

In order to study data at a national level, which is available for more years, the paper opted for the method of panel data analysis. Panel analysis takes into account the diversity of individual research units whilethe combination of time and cross-sectional dimensions allows for more informative, more various and less collinear variables, providingmore degrees of freedom and better efficiency and reliability of the estimators (Baltagi, 2005). The characteristics of the dependent variable determine which is more appropriate to use: the static or the dynamic panel model. The dynamic models result in unbiased and consistent estimators only with a larger number of cross-section units (N) and time periods (T) (Škrabić, 2009). Sincethis study included only 28 EU countries and the longest time series for some variables extends up to 8 periods, this dataset is not appropriate to apply the dynamic panel analysis. Therefore, the paper used only static models.

The data for the analysis was collected from secondary sources. They refer to the 28 members of the European Union. The dependent variable is the level of innovativeness measured by the business sophistication and innovation index (BS&I). The current way of measuring BS&I is available only for the last 8 consecutive periods through the data platform (WEF, 2013). Most of the explanatory variablesdata was extracted from web-based data platforms (Eurostat, n.d.; Worldbank, n.d.). In addition, data on PP and PPI was collected from the reports of the European Commission on public procurement indicators (EC, 2009, 2012) and from the global competitiveness report database platform (WEF, 2013). European Commission reports do not include data for Croatia (which joined the EU in 2013) and therefore our estimate for Croatia used the annual reports of the Croatian Ministry of Economy (MINGORP, 2013) and data on the Croatian GDP (Worldbank, n.d.). The advantage of secondary sources is the fact that their data collection methodology is known and relies upon a theoretical basis, thus producing comparable data for different countries. The disadvantage of secondary sources is practical field collection of the data by many national organizations, where there may be different interpretations of the otherwise same guidelines. Also, sometimes there are countries that lack data for certain time periods,which creates anunbalanced panel. Even though our data is unbalanced, the Stata software used in the analysis has procedures that correct the estimates for unbalanced data.