The firm’s ability to innovate is one of the usual suspects for explaining differences in firm performance according to a strong and diversified theoretical framework [1,2]. Innovation facilitates the high rate of growth of ‘‘superstars’’ as well as the establishment and continued existence of profitable companies that do not seek to become large enterprises [3]. Understanding the diversity that exists within the population of innovative firms is essential to develop appropriate innovation policies. Our study explored the diversity of innovation patterns among Norwegian firms and identified typical approaches to innovation, which connect innovation inputs and outputs at the firm level.

The mechanisms linking R&D, innovation success, and firm performance are largely indebted to the Schumpeterian endogenous growth representation, according to which firms strive to innovate so that they can enjoy monopoly rents [4]. The forward-looking firm makes a decision about its level of research input, based on the expected returns to R&D (in terms of sales or directly in terms of profits), which affects the stochastic innovation process. Innovation success, in turn, automatically raises the firm’s profitability or productivity level [5,6]. Such stochastic and optimizing representation has, however, been challenged by models in which agents constrained by bounded rationality search for more productive techniques in an uncertain environment, where the impact of innovation on firm growth is itself random [7]. In such a framework, firms are heterogeneous in their ability to innovate, not only because of their financial resources, but also because they differ in terms of their ability to exploit technological opportunities. Path dependency explains the concentration of many innovations in the hands of a limited number of firms [1], while there may also be growth pattern heterogeneity for the same levels of R&D due to the uncertain nature of the R&D process [8]. Even among successful innovators, heterogeneity persists: while innovators are likely to enjoy superior employment growth compared with non-innovators, the bulk of this differential derives from the exceptional job creation activities of a few firms [9].

If policy-makers are willing to help different firms (incumbents or entrants), a first way to group the target firms is by the type of products and processes they deal with, which, in turn, roughly defines the economic sector to which the firms belong. At high levels of aggregation, product-based classifications of sectors such as the NACE system have often been considered impractical for understanding the sectoral dynamics of innovation. Therefore, other classifications have been suggested for this purpose, which use a finer disaggregation level as a basis for new definitions of economic sectors. Pavitt [10] proposed a four-sector taxonomy based on size, innovation patterns, and sources of innovation: scale-intensive, supplier-dominated, science-based, and specialized supplier. Miozzo and Soete [11] proposed removing services from the supplier-dominated category in Pavitt’s original classification and suggested four additional categories: supplier-dominated services, physical network services, information network services, and knowledge-intensive business services. This led to an eight-fold taxonomy including four manufacturing and four service sectors; the taxonomy was later subjected to further aggregation by Castellacci [12].

However, these taxonomies have still grouped data at the level of industry rather than that of firms. Such a choice ignores the fact that firms in the same industry may have a very different technological base. This issue was raised by Archibugi [13], who said that “[h]opefully, over the next few years more statistical and econometric work will be carried out to group firms, as opposed to industries, into the taxonomic categories [.] according to their intrinsic characteristics such as the rate and direction of technical change and their sources of innovation” ([13] p. 420). Their hope was partially misplaced, since data limitations have often constrained the researchers in innovation studies into using output-based sectoral definitions. In our study, we used firm-level data from the Innovation Survey conducted in Norway in 2018 to identify various approaches to innovation. Drawing on De Jong and Marsili [14] and Leiponen and Drejer [15], we employed a factor analysis to reveal typical patterns of innovation behavior by analyzing correlations in the answers to the survey. Unlike in previous studies, we did not aim to label each firm as having one specific approach to innovation, but we allowed for the coexistence of several approaches to innovation within the same firm. The eleven innovation patterns we identified are therefore eleven different, but not exclusive, ways for a firm to be innovative.

We took stock of the findings by Baregheh et al. [16] who collected different definitions of innovation from various disciplinary literature and, following a content analysis, proposed that “Innovation is the multi-stage process whereby organizations transform ideas into new/improved products, service or processes, in order to advance, compete and differentiate themselves successfully in their marketplace” ([16] p. 1334). This definition helps us to reaffirm, on one hand, that innovation is a process rather than a discrete act and, on the other hand, that the presence of an aim is necessary to reconnect the rhetoric
of innovation to its strategic contexts. Our analysis uncovered the innovation patterns of Norwegian firms by studying not only the variables associated with innovation inputs and outputs, but also variables describing the goals and hindrances in the innovation process, as emerging from the answers to the 2018 Norwegian Innovation Survey.

Furthermore, Section 2 describes the existing literature on which our study is based. Section 3 explains how we constructed and used variables for our factor analysis. Section 4 describes the results of the analysis, while Section 5 provides a discussion of the results. Section 6 presents our conclusions.