The role of public procurement as innovation lever: evidence from Italian manufacturing firms

ABSTRACT

The study focuses on the impact exerted on private R&D expenditures by regular and innovative public procurement when taken in combination or insolation with supply-push measures. The econometric analysis relies on a pulled sample of 4206 Italian manufacturing firms observed between 2010 and 2014. The empirical exercise confirms previous evidences on the relevance of technology-push instruments in sustaining firms’ innovativeness. On the contrary, the ability of public procurement activities in shaping innovative investments is found to depend on a number of instances related to: (i) the adoption of contemporaneous supply side measures; (ii) the inclusion of innovative demand in procurement contracts. The analysis provides important suggestions with respect to the potential effectiveness of demand-side tools when implemented in weak administrative and innovation systems, as in the Italian case. Moreover, it is shown that the design of the policy mix matters, and its effectiveness improves when demand-side and supply-side instruments are jointly implemented.

KEYWORDS:

• Demand-pull policies
• public procurement
• policy-mix
• non-parametric analysis

SUBJECT CLASSIFICATION CODES:

• H57
• O25
• O38

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 This paper has been developed under the research agreement between SI.CAMERA – SISTEMA CAMERALE SERVIZI SCRL and the Department of Economics of Roma Tre University. The funding from MUR-PRIN GRANT NUMBER: 20177J2LS9_003 is also gratefully acknowledged. This paper represents the opinions of the authors. Any errors are the fault of the authors.

2 We acknowledge the key role played by services in the furniture of the most advanced goods, especially with regard to the health sector. However, although their exclusion from the sample might make our findings less generalizable, the choice of focusing on industrial firms might allow for a better comparability of our results with existing studies, as most of them are concentrated on manufacturing. Secondly, given the semi-parametric framework here adopted, narrowing the exercise on units which are more similar one each other in terms of sectoral characteristics helps in smoothing selection-bias related issues.

3 Data available at the web site: https://stats.oecd.org/index.aspx?r=171728#

4 More in general, the adoption of a static efficiency attitude in procurement activities implies that the main objective of procuring agencies is to take advantage of the price competition between suppliers in order to purchase goods/services at the lowest possible cost, within the existing basket of technologies. Conversely, adopting dynamic-efficiency criteria in procurement tenders means that public demand is intentionally oriented to stimulate both the creation and the dissemination of new technologies and to spur dynamic competition between suppliers. Hence, dynamic efficient oriented public procurement pays specific attention to the innovative content of the acquired goods or services and it is not merely guided by cost related considerations.

5 For a detailed description see EC (2018)

6 Available at the webpage: https://ec.europa.eu/regional_policy/sources/policy/how/improving-investment/public-procurement/study/country_profile/it.pdf:

7 Law Decree 50/2016, Tree-year plan 2017–2019 of AgID (Digital Italy Agency), the 2018 Framework Agreement for Growth and Digital Citizenship in line with 2014 European directives (23/EU, 24/EU and 25/EU).

8 ‘More specifically, the ‘other’ innovation policy tools we refer in the text basically concern supply-push innovation policies. With respect to the Italian context and the time-window considered (2010–2014), we list the main SP policies that were in force during the period under scrutiny. The majority of these measures were financed by national founds (FIT, FAR, FAS, FNI, FRIM, Fondo Centrale di Garanzia, Fondo Italiano di Investimento) consisting in: (i) non-repayable and soft loans for the acquisition of capital goods, (ii) public guarantees on bank loans and (iii) tax credits on increases in R&D investment.

9 To identify IPP in CIS questions, we rely on innovation activities undertaken by firms ‘as part of procurement contract’, thus regardless innovation is intended as ‘formal’ requirement or not. This point is deeply discussed by Edquist (2015) who argues that, according to the different forms whereby IPP might be set up, contracted innovations might be intended both in terms of ‘formal’ or ‘informal’ requirement to innovate.

10 451 innovative manufacturing firms have been dropped from the analysis because of missing balance sheet data.

11 Unfortunately, given the lack of more detailed information about the nature of the public aid received from innovative firms, we cannot distinguish between different types of instruments, i.e. public R&D subsidies, tax credits or loans.

12 We acknowledge the limited relevance of the sample of firms involved in IPP. The results presented in this paper should be intended as a first attempt to provide evidence on this specific type of PP contracts, which has been rarely studied in a quantitative policy evaluation context of analysis.

13 The implications of CIA are discussed in Cerulli (2015).

14 We acknowledge the difficulty to control for every source of selection bias with considering the available data. Actually, CIS questions do not inform about firms’ strategic orientations, intangible assets, tacit knowledge, and other relevant characteristics. However, by merging the CIS surveys with AIDA balance sheet data, we attempt to capture the main dimensions reflecting firms’ propensity to engage in PP activities (see Table 2) by looking both at firms’ structural characteristics and sectoral aspects. Surely, this might not be sufficient to entirely cope with selection-bias issues, however we believe that our list of covariates allows to control for the major sources of them.

15 Other measures typically used for financial constraints (as for instance indebtedness represented by bank debts) are not adopted because they are missing for many observations.

16 With respect to high and low-tech industry, we have followed the OECD ISIC Rev. 3 technology intensity definition of manufacturing industries. In particular, we have grouped together firms in high and medium-high technology sectors into the unique category of high-tech industry; likewise, firms in low- and medium-low technology sectors have been grouped together in low-tech industry.

17 The teffects psmatch command in Stata14 is used to calculate ATT, since it estimates standard errors adjusted for the first-step estimation of propensity scores, as suggested by Abadie and Imbens (2016). The balancing property of the propensity score is tested using the Becker and Ichino (2002) user-written Stata command pscore and is always satisfied.