Procuring NGA infrastructure: The performance of EMAT auctions in Italy

Highlights

• Policies for NGA roll-out build on very elaborate public procurement schemes.

• In first-price, sealed-bid auctions EMAT can elicit vibrant bidding competition.

• EMAT rules can effectively procure quality-intensive and uncertain high-tech infrastructure.

• EMAT rules can enable multidimensional procurement agendas.

• Italian procurement strategies were effectively complemented by pro-active industrial policy.

Abstract

Italy was a patent laggard in the market roll-out of NGA networks. In 2015, the Italian Government introduced the “ultrabroadband strategy”, an ambitious policy agenda addressing the NGA market failures. This paper assesses the procurement policy assigning public concessions to build and manage passive infrastructure in “NGA white areas”: in detail, the techno-economic properties and outcomes of the auctions are analyzed, both theoretically and empirically. Because the applicable procurement law was the EU one, the EMAT scoring rules were mandatory. Relevant findings stand out, which confirm the received wisdom on first-price auctions, and extend it to high-tech sectors, where EMAT remains under-researched. First, the concession auctions unleashed vibrant bidding competition in many white areas, which enabled substantial budget savings. Second, the quality scoring rules provided adequate flexibility to prioritize the procurement of the most innovative and pro-competitive architectures (FTTH/B/P), despite the presence of high uncertainty. Third, the EMAT framework catered for a multidimensional procurement agenda; the latter targeted efficient and innovative procurement, universal service and higher market competition, while accommodating the ambitious industrial policy of the Government. Evidence of a potential trade-off between the plan implementation delay and its innovativeness emerged; such a conflict was eventually exacerbated by the bandwidth shortages generated by the COVID-19 crisis. By providing the very first study on EMAT auctions for broadband, this work stimulates future comparative research on additional EU member States.

Introduction

The slow and difficult transition to next generation access (henceforth, NGA) broadband of most European countries has been pointed out early on, together with its common causes, including market and policy failures (Briglauer & Gugler, 2013). Italy stands out in the group of the laggard EU member States: its market registered a delayed roll-out of networks – initially with first-generation broadband, then with NGA (Bourreau et al., 2010; Nucciarelli et al., 2013; Matteucci, 2014, 2019; Quaglione et al., 2020). Moreover, the supply-side deficit was aggravated by the unsatisfactory rate of broadband subscriptions.

Since 2015, the Italian policy-maker undertook a new type of “vertical” industrial policies to solve the NGA market failures, and the project was labelled as the “ultrabroadband strategy” (PCM, 2015a; 2015b). Following the taxonomy of Falch and Henten (2018), such a strategy matches the case of “broadband developmental” policies, and those improving infrastructural competition. The ultrabroadband strategy was not the first initiative. In fact, the very first public plan for NGA was formally issued by the Italian Government in 2011 (Matteucci, 2014; MISE, 2011a), and was authorized as State Aid (henceforth, SA) measure n. 34199 (EC, 2012a): we call it “the first NGA measure”. However, that plan lacked a specific budget, and was eventually implemented only in those Southern regions possessing unspent cohesion policy funds (Matteucci, 2020a), to cover urban areas. Indeed, the ultrabroadband strategy announced in March 2015 is the first credible country-wide Government project backed by adequate funding – henceforth, we call it the “2015 strategy”. In early 2016, the part of the strategy targeting the NGA “white areas”¹ was finalized; then, in June 2016, it was approved by the European Commission (SA measure n. 41647; EC, 2016a). This policy case is multi-dimensional and includes, besides issues of auctions, public procurement and antitrust (such as State aid - henceforth, SA), also regulation and industrial policy.

Historically, Italy did not develop cable TV. For a long time, the new telecoms entrants had to rely on the copper local loop of the incumbent, TIM.² However, this bottleneck infrastructure was subject to a defective regulatory regime based on a weak model of functional separation (Nucciarelli & Sadowski, 2010): henceforth, TIM enjoyed persistent market power, prone to abuses (Gerli & Pontarollo, 2013). Moreover, from end-2012 to 2016, the incumbent's and the entrants' private NGA investments were focused on the FTTC architecture³: thereby, they all depended on the incumbent's local loop, and showed wasteful overlapping duplications (AGCM-ACGOM, 2014; Caio et al., 2014). In 2016, a fundamental market change occurred. A newco - ENEL Open Fiber Spa - entered the telecoms market and formally broke the local loop monopoly by announcing the construction of an alternative country-wide infrastructure, based on full fiber optic cables (FTTH).⁴ ENEL Open Fiber Spa adopted the “wholesale-only” business model, and initiated an ambitious private investment plan targeting the most important urban areas (now totaling 271 towns).⁵ Furthermore, the newco participated in the Government's auctions supporting the NGA roll-out in “white” areas, and won all the parcels. The newco is jointly owned by two major State-invested and controlled companies – the electricity incumbent ENEL and the long term investment bank Cassa Depositi and Prestiti (henceforth, CDP). Henceforth, the ENEL Open Fiber Spa's project also signals the resurgence of industrial policy for telecoms in Italy (Matteucci, 2020b).

These events have rapidly modified the Italian NGA supply-side and market structure. By looking at the latest releases of the DESI index-Connectivity⁶ and the FTTH Council Europe evidence,⁷ we see that, while Italy remains a laggard country for the actual adoption of NGA and ICT services, it has rapidly caught up as far as the NGA roll-out and coverage are concerned. By September 2019, Italy has become the fifth top country in Europe in terms of absolute number of FTTH/B sockets deployed (equal to 11,150k).⁸ Henceforth, an investigation of the NGA policy-side is called for.

We carry out an evaluation of the SA measure n. 41647 (PCM, 2015a), to study its techno-economic properties and systemic effects. As noticed by Briglauer et al. (2014), our knowledge of the economic performance of broadband policies remains scarce. This is particularly true of industrial policies supporting networks roll-out and universal coverage: to the best of our knowledge, there are only three works. Briglauer et al. (2019) evaluate the SA policy supporting the coverage of first generation broadband (henceforth, FGB) in Bavaria, Germany; Gerli et al. (2020) that of NGA broadband in UK; Matteucci (2019) that of FGB in Italy.

Owing to its early stage of implementation, the final impact assessment of the SA measure n. 41647 is not yet feasible. As an alternative, we focus on the Government's auctions procuring NGA infrastructure for covering market failing areas. In the EU, more than elsewhere, broadband policy has been strictly connected with public procurement policy, besides SA policy. In fact, the EU constitutional Treaties rest on the project of building the Internal (or Common) Market, and the latter requires that member States and public procurers do not discriminate supplying firms on grounds of nationality. In turn, the EU public procurement policy was reformed by two waves of Directives, the 2004 and the 2014 packages, which further harmonised national norms.

Hence, a main research question of our work is whether the current EU-driven framework of public procurement applicable to public works is adequate for implementing NGA policies. Such a question is warranted for various reasons. First, NGA policy procures an infrastructure that is technologically more complex and uncertain (about its quality) than traditional public works (roadwork, buildings, etc), or frequently procured supplies (stationery). Second, EU broadband policy (EC, 2013) encompasses multiple goals: among these, solving market failures and guaranteeing “universal service” for vulnerable consumers and areas, procuring future-proof infrastructure for the “Gigabit society”, promoting infrastructure-based market competition, enhancing the competitiveness of downstream sectors by providing reliable connectivity services, etc. Third, broadband policy should also ensure the efficient and effective usage of taxpayers' money – and this is the most typical mandate of public procurement. Hence, one could legitimately doubt whether these numerous and ambitious policy targets could be jointly achieved with the ‘simple’ policy tools conventionally used to procure items such as road construction or stationery. Indeed, EU policies may present conflicting sides (e.g., Colomb & Santinha, 2014; Wishlade, 2012), and new challenges may emerge after reforms - such as the recent packages on EU public procurement.

While studying the Italian auctions procuring NGA infrastructure, we focus on their key awarding mechanism – the method of the “economically most advantageous tender” (EMAT, or MEAT). By the EU law, this method is mandated for the auction adjudication of public works whose price surpasses stated threshold values. The salience of this research topic is high: so far, despite the fact that EMAT had to be used in many EU broadband SA measures, it has not been studied, and its actual performance and capacity to deliver the intended policy goals remain completely undocumented. To the best of our knowledge, our contribution is the first in the literature. Regarding EMAT, this paper accomplishes the following key tasks. First, it reviews the insights coming from the standard “first-price”, “sealed-bid” auctions (e.g. Füllbrunn & Neugebauer, 2013), and combines them with those available on EMAT scoring rules (from pioneer to recent papers, e.g. Che, 1993; Bergman & Lundberg, 2013), across the different sectors of application (Hyytinen et al., 2018; Lundberg & Bergman, 2017); then, it conceptualizes them as testable hypotheses. Second, it builds an original theoretical and conceptual framework enabling the empirical assessment of the Italian auctions procuring NGA infrastructure, and retrieves the relevant evidence, so far unreleased. Third, by using this empirical evidence and focusing on the quality-price mix and the trade-offs embedded in the EMAT scoring rules, it investigates what kinds of policy targets were eventually delivered by the Italian NGA procurement. This test extends the scanty empirical literature on EMAT auctions to new sectors of application, such as the procurement of high-tech and quality-uncertain infrastructure. Finally, the paper explores preliminarily the factors harming the progress of the Italian NGA strategy for white areas, due to be completed by 2020.

Section 2 presents the EU public procurement law, the literature review, and the EMAT functioning. Section 3 introduces the 2015 Italian strategy supporting the NGA roll-out, and its SA measure n. 41647; then it stylizes the Italian auction design and the procurement strategy; finally, it presents the research method and hypotheses. Section 4 presents the empirical analysis, which studies the auction outcomes based on traditional price indicators. Section 5 refines the empirical strategy, adds the quality dimension and discusses findings specific to EMAT multi-dimensional agendas. Thereby, section 4 Empirical analysis: main findings, 5 Refinements and discussion perform an in-itinere assessment of the Italian NGA strategy. Section 6 concludes, distilling the main policy implications and dilemmas, for Italy and the other EU member States.