**2. Co-innovation and the role of university**

In analyzing the role of universities as innovation centers, regional development theories underline the importance of agglomeration economies: firms cluster within a territory where they share knowledge, production inputs, labor market regulations, and production processes. The well-known Triple Helix model by Etzkowitz, Leydesdorff, and others [19] articulates the development of regional hi-tech industry clusters as a networked combination of university, industry, and government organizations collaborating in the promotion of innovation. In these configurations, universities act as intermediaries to transform knowledge into networks of multiple agents [6, 20]—i.e., companies, institutions, research centers, foundations, banks, and international organizations. Carayannis et al. [5] stress that the university is a new type of organization—"capable of higher-order learning, a type of open, highly complex and non-linear knowledge production system" (p. 152). This endogenous role of universities emerges out of recursive linkages in UICs that follow historical trajectories of spatially localized learning.

A key issue of this debate concerns knowledge generation and exploitation within proximity contexts [8, 21]. Productivity gains from UICs are highly localized, with companies situated near universities able to introduce innovations at a faster pace than rivals located elsewhere [22]. Assessing the efficacy of university in regional innovation, the size of the institutions involved is not a key factor [3] nor is the university research intensity or quality [23]. Universities' mission overload may hinder local firms' ability to cooperate or force them to look beyond the region for other suitable universities to interact with.

To understand interdependencies in UICs, business studies draw on the innovation ecosystem theory to reconstruct the variable set of actors, activities, (digital) artifacts, and the complementary and substitute relations, which are key for local innovative performance [14]. Granstrand and Holgersson [24] emphasize that an innovation ecosystem is different from innovation systems broadly considered, in that participants exhibit concomitant collaborative and competitive behaviors. While innovation ecosystems allow for cross-sectoral and cross-regional examination of innovation activities, other conceptualizations are often based on geographical boundaries, labeled using constructs such as national or regional innovation systems [24, 25]. Furthermore, a business ecosystem finds its roots in the idea of value networks of companies that combine skills and assets [6, 24]. In this strand, technology business studies focus on platforms, where data availability builds the infrastructure that facilitates entrepreneurship, new market opportunities, and social impact on businesses and society [26].

Considering the interactions between companies and universities, business studies examine knowledge-based value creation processes leading to invention and commercialization [6, 7, 27, 28]. Knowledge ecosystems aim to create the conditions for the exploitation and the appropriation of the value associated with knowledge, whereas business ecosystems involve the broader scope of knowledge exploitation or the process of invention-to-commercialization [6, 7, 27, 28]. Knowledge ecosystems work in precompetitive contexts, where physical proximity to universities and large firms with established Research and Development (R&D) departments have a positive influence on the focal actor's innovative output [6, 29].

Yet, Stam [4] points out the lacking causal depth and evidence base in ecosystem studies, whereas Clarysse et al. [6] highlight the implicit assumption that business ecosystems are the consequence of setting up a knowledge ecosystem, with universities, public research centers, and businesses that cooperate in early stages of knowledge creation. But, how actors participate in knowledge exploration, how they coordinate actions, and what effects emerge out the cooperation between universities and companies remain still under-researched.

Thus, this chapter seeks to fill this gap drawing on Adner's [30] distinction between the ecosystem as affiliation—that examines interdependencies like other well-established approaches of networks, platforms, markets with multidimensional players—and the ecosystem as structure—that interprets the ecosystem as a variable configuration of activities connected by a value creation strategy. The ecosystem as affiliation focuses on the access to the system and the number of partners, the network density, and actor centrality to understanding growth strategies for an actor or platform to assume weight and power. The ecosystem as structure assesses a value creation strategy based on the need for actors to collaborate. Actor alignment not only mirrors mutually compatible incentives and motivations, but it also verifies the conditions for which partners interpret the multilateral configuration of relationships that cannot be broken down into an aggregation of bilateral interactions [30]. If actors do not express the need to align for sharing strategies, because they are not crucial partners in the value creation process, or because their relations do not question the current alignment, the value-added advantage associated with an ecosystem is not necessarily relevant.

The methodological implication of this framework is the need to explore the rationale and behaviors of participants to reconstruct the nonlinear, multilayer, and recursive causality associated with ecosystem-based complementary and competitive relationships. The interactions in cooperative, competitive, or antagonist relationships get usually modeled through network links between nodes with attractive and repulsive couplings or positive/negative ties. These structures have layers in addition to nodes and links. Thus, a node in a layer (composing a single network) can get linked to any node in any other layer. Layers represent aspects or features that characterize the nodes or the links that belong to that layer. Links can get partitioned into intralayer links (e.g., links that connect nodes set in the same layer) and interlayer links, which tie nodes set in different layers. Furthermore, network analysis in ecosystem studies involves either whole-network analysis at the macro-level or ego-network analysis at the micro-level. By contrast, macro-level and micro-level networks may include multilevel nodes and interlevel edges. For example, a social network of researchers (micro-level) and a resource-exchange network between laboratories (macro-level) to which the researchers belong constitute a multilevel network with two levels. Few studies have investigated network community analysis at the meso-level [31]. Building upon these theoretical premises, the next section presents the research design examining the knowledge-based value creation process the University of Naples and technology multinationals have pursued through digital training partnerships and what factors have contributed to its growth.
