Research data can be, and often is, viewed as a public good. Research is often at least a partly publicly funded activity and the results are expected to be made publicly available and in the service of public interests.

A public good is non-excludable (meaning that no one can be excluded from using or benefiting from it), and non-rivalrous (meaning that its availability is not diminished by its use) (Ostrom & Ostrom, 2019; Samuelson, 1954). Examples of public goods are basic infrastructure such as sanitation, street lighting, and public education. Typically, governments are responsible for the provision of public goods, and they are funded by taxes. An important function of public goods is to correct for market failures. As an example, if education were to become a purely private good, many would lose access to schools due to a lack of affordability, and inequalities would likely grow due to reduced opportunities for social-economic mobility through education. Research data can be viewed as a public good when it is in part publicly funded and expected to be available to the public. Further, research data can be said to be a digital public good as most of the data produced today is in a digital format (Kleppner & Sharp, 2009).

Research data as a public good is an important part of the Open Science (OS) movement (Vicente-Saez & Martinez-Fuentes, 2018), which has the objective of providing free-of-charge and unrestricted access to publicly funded research. Open Science comprises several research outputs beyond data, such as open academic articles, open software, and open educational materials. The OS movement also has implications for the research process itself, shifting traditional closed practices toward open ones – for example, open peer review and citizen science. The implementation of OS has been gaining momentum through research funder mandates (e.g., Directorate-General for Research and Innovation, 2021). Several principles are highlighted in OS, among them transparency, equality of opportunity in accessing and taking part in science, integrity when carrying out science, inclusion of marginalized communities and geographical regions, and increased collaboration in addressing society’s complex problems, such as social inequality and environmental degradation (UNESCO, 2022). More recently, a focal point of the Open Science movement has been on how to make research data as open as possible (Landi et al., 2020). The “as open as possible” approach to research data recognizes the need to balance openness with other considerations, such as regulatory frameworks. One example is the General Data Protection Regulation (GDPR) wherein the privacy of individuals is heavily protected, and limitations are imposed on the processing of data containing information on individuals (e.g., Mondschein & Monda, 2019).

Many benefits are associated with research data as a public good. For one, the re-use potential of research data can be high. When properly managed, research data can be used for multiple purposes in different projects over time, especially when combined with other data. One could argue that re-using data is a sustainability issue because it may be unfeasible for the research enterprise to collect or generate new data every time a new research project is started. Following this line of argumentation, it is useful for research data to be put to new uses so that its value is extracted to the highest possible extent in service of advancing knowledge. Also, data re-use may help diminish undesirable consequences to the environment. One such undesirable consequence can be, for example, carbon intensive air travel by a research team for data collection purposes.

Other benefits of research data as a public good is that it can inform the design and evaluation of public policies (e.g., Crato & Paruolo, 2019), assist in the fight against pandemics and prevention of diseases (Nature, 2021), help in the monitoring, protection and restoration of the natural environment (Urbano et al., 2023), contribute to the development of high-quality educational resources (Coughlan, 2020), and in general help promote sustainable development (Pappas et al., 2018).

At the same time, research data has a clear market value and can be viewed as a private asset because research data can be privately owned and used commercially. Research data is a versatile resource that can serve as an input upon which products or services are built, especially in light of emerging technologies that can automate the compilation and analysis of data.

Research data can be classified as an intangible asset, which is a non-physical resource that can generate earnings for an organization. Intangible assets can for instance consist of an organization’s brand, patents, intellectual capital, and customer relationships (Buonomo et al., 2020).

Research data as an asset can be understood within the Resource Based View (RBV). The RBV proposes that the bundle of resources and capabilities of an organization constitutes its source of competitive advantage, and when these resources and capabilities are valuable, rare, imperfectly imitable, and not substitutable, they can be the source of sustained competitive advantage (Barney, 1991; Barney et al., 2001; Wernerfelt, 1984). It can be difficult to estimate the value of research data as an asset, as well as its role in providing organizations with sustained competitive advantage. This is partly because research data can vary in value throughout its lifecycle. For instance, raw data can have a wide range of applicability, but likely a lower value due to the need for clarifying its use cases. Processed and curated data can have a narrow range of applicability, but likely a higher value due to already identified use cases. We argue that not only economic benefits can be expected from the use of research data as a private asset but also gains in competitiveness.

Beyond arguments on economic returns and competitive advantage, one could argue that research data ought to be governed as a private asset because of the need to cover the costs involved in generating and maintaining high-quality research data, and because of the under-funding of this activity. Research data generation and upkeep can involve many hours of labour and sometimes also the use of expensive equipment. Activities related to research data management and stewardship are typically under-funded (Mons, 2020). This is often the case in the planning and execution of research projects, where the funding is not unlimited and typically must be prioritized for activities other than data management (such as the production of academic publications or the deployment of a communication strategy for the project). Therefore, the argument continues, RPOs may need to consider commercialization prospects for their research data if they intend to govern data in a financially sustainable way.

Additionally, there is a cost in making research data FAIR. The acronym FAIR stands for Findable, Accessible, Interoperable and Reusable, and managing data according to the FAIR principles means handling and documenting data in a way that makes it understandable and reusable (Wilkinson et al., 2016). Before becoming FAIR, research data needs to be enriched with contextual information that allows others to find and understand it, as well as being supported by technical infrastructures and policies that allow the exchange of data within and across RPOs and other organizations. All of this means that there is a cost for RPOs to “FAIRify” data.

Another cost is connected to the evaluation of the re-usability potential of research data. The value of data beyond its first use in a research project needs to be discussed and assessed. Research data can be of a highly contextualized nature where their generation and processing happen with very specific applications in mind. This contextualized data allows researchers to draw insights that are highly relevant to the research questions being explored in a project, but, at the same time, this may reduce the transferability of the data to other contexts and applications. For example, interview data is typically highly contextual and difficult to transfer to other contexts. While preserving this type of data for reproduction purposes is important, its re-application to other contexts and potential commercial applications is uncertain. Not to mention that regulations, such as GDPR, can substantially limit the reusability of this type of data. All of this means that an effort has to be made by RPOs in evaluating the re-usability potential of their research data. This represents another cost for RPOs, as substantial work may be required to make these evaluations and to select which data is worth preserving and for what ends.

Considering the use of research data for commercial ends may be interpreted as taking precedent over open research data even in publicly funded projects. For instance, the European Commission states that beneficiaries are required to adequately protect their results and consider possible prospects for their commercial exploitation (Directorate-General for Research and Innovation, 2022). The Horizon Europe program states that beneficiaries need to make it viable for research results to be commercially exploited through, for instance, the transfer and licensing of results (European Commission, 2022). These statements highlight the commercial value of research outputs including research data.

It is important to note that these discussions go beyond the research data that underpins claims in scientific publications. Once research results are disseminated, the accompanying data typically needs to be made available for verification purposes (except when there are legitimate reasons for not doing so). This is not only a compliance issue, but also a best practice in research. Our discussion concerns the overall research data generated by individual projects and the body of data accumulated from projects over time.

Currently, the costs for generating, managing, and preserving research data are high and rarely sufficiently accounted for. This can result in inefficiencies, unfulfilled potential, unsustainable financial practices, and missed opportunities. We also suggest that the protection of research data for commercial exploitation may take precedent over its open dissemination. Consequently, within the view of research data as an asset, the strategic management of research data by RPOs – in a similar way to how they manage other intangible assets – can prove to be very important. This is particularly important under increasingly competitive conditions for obtaining research funding (e.g., Slaughter & Rhoades, 2009).

We believe most RPOs rely on the overarching principle “as open as possible, as closed as necessary” to navigate the tension between the market and non-market values of the research data they hold. However, we argue that this principle does not provide sufficient guidance for RPOs. Neither does it converge with the growing expectation that research results ought to benefit society via market mechanisms, nor does it fit perfectly with the requirements from the European Commission regarding the need to protect research results. More guidance is needed for RPOs to move closer to an integrated strategy to research data governance.

Although the principle “as open as possible, as closed as necessary” provides an overall orientation for RPOs – that is, it prescribes a primary course of action “share research data as open as possible” followed by the complementary action of “protect research data as necessary” – it does not give RPOs concrete advice on how to achieve a balance between open and closed data. Other related initiatives, such as the DORA declaration,¹ also do not provide sufficient guidance for RPOs in this respect. The DORA declaration aims to recognize a wider range of research outputs beyond scientific publications. This includes, for instance, the production of datasets and software. However, despite being an important driver of new research evaluation practices, the declaration does not guide RPOs in how to strike a balance between the open dissemination and the protection of research data.

Therefore, more guidance is needed in this respect. With this purpose in mind, we suggest that the adoption of the Resource Based View and Open Science can provide insights for the development of a data governance model that acknowledges research data as a public good and as a private asset. We argue that RBV and OS can be seen in connection with Scheiner’s (2020) framework, which deals with opposing perspectives that carry an inherent tension between them. Scheiner proposes four approaches in this regard: (1) to treat the perspectives as lacking a common ground and contradictory, where one perspective is chosen over the other (either-or); (2) to treat the perspectives as in constant tension, preserving the paradoxes between them (back-and-forth); (3) to treat the perspectives as independent and yet complementary, placing them in dialog with each other (both-and); and (4) to treat the perspectives as blending into each other, merging them into a synthesis (more-than).

Each of Scheiner’s four approaches has merit and can help solving different types of conflict. For our purposes, we believe Scheiner’s third approach (i.e., both-and) would suit RPOs best as this approach emphasizes complementarities between research data as a public good (as part of Open Science) and as a private asset (as part of the Resource Based View). We argue that the remaining alternatives would not suit RPOs as well for the following reasons:

Therefore, we believe that Scheiner’s third approach (i.e., both-and) suits RPOs best. This approach represents a step forward for how RPOs can find ways to deal with the market and non-markets values of research data, in a complementary fashion. We illustrate this approach in Figure 1, and then we explore the interplay between RBV and OS in more detail.

A foundational element of the Resource Based View is the strategic management of organizational assets (Wernerfelt, 1984), which includes research data. RPOs can strategically manage their research data assets, exchange them internally and externally, and establish collaborations based on these assets in mutually benefiting ways. Governing research data strategically contributes to more sharing and re-use of data internally and across organizations, benefiting and promoting Open Science.

Strategically managing research data involves taking steps to ensure that this activity is funded in a sustainable way, as this can enhance the long-term preservation and availability of data. This is beneficial for Open Science as more research data is likely to be findable, accessible, interoperable, and re-usable (FAIR) when sufficient resources are directed to this activity. RPOs can, for instance, seek revenue streams from selected data assets via direct selling or licensing, and such revenue streams can be used to support FAIR data management. Our main argument, in this regard, is that the commercialization of research data can be used to sustainably finance FAIR data management at research performing organizations, thus contributing to Open Science.

Another important aspect of the Resource Based View concerns resource allocation (Peteraf, 2005). It is often the case that resources are limited and need to be prioritized. RPOs need to allocate resources efficiently, not least in regard to research data management. Treating research data as an asset can help RPOs make decisions regarding resource allocation. For instance, RPOs may choose to prioritize the preservation of higher-value data assets, followed by lower-value ones. Note that research data can be valuable for both market and non-market applications and that it is important for RPOs to assess their data in light of both purposes. In short, we argue that RPOs can benefit from assessing the market and non-market values of their data assets and from using this assessment to make resource-allocation decisions. This is beneficial for Open Science because higher-value research data assets are more likely to be re-used than lower-value ones.

Lastly, following the argument that strategic partnerships can be the source of competitive advantages (Dyer & Singh, 1998), we suggest that RPOs can benefit from forming partnerships with organizations holding complementary data assets. This way, RPOs can achieve mutually beneficial exchanges and jointly held competitive advantages. This is beneficial for Open Science as RPOs can learn about collaboration opportunities based on research data and share resources and capabilities across organizations, thus avoiding the duplication of efforts and enhancing overall efficiency. All of which are key goals of Open Science.

Open Science promotes the sharing of research data and related materials, such as methodologies and data management plans (Staunton et al., 2021). The increased availability of these outputs can help RPOs access resources that are not held by any individual organization in isolation. RPOs can then create their own mix of internally and externally held research data, which can provide them with unique combinations of resources and competitive advantages. Open Science – via the increased sharing of research data – can then inform the Resource Based View by making available a larger resource pool to RPOs.

As Open Science promotes the wide distribution of knowledge, it creates appropriate conditions for organizational learning and innovation (Directorate-General for Research and Innovation, 2016). RPOs can stay up to date with the latest scientific and technological advancements and achieve competitive advantages based on where they direct their learning efforts, rather than on information withholding. Open Science can contribute to more efficient resource use and innovation, as RPOs have more knowledge at their disposal and a reduced need to duplicate their efforts in searching for scientific and technological breakthroughs.

Open Science enhances responsible research practices, such as transparency and integrity, and increased collaboration in addressing societal problems (UNESCO, 2022). This can inform the Resource Based View as the possession of ethical values and practices is, by itself, an organizational resource. This resource contributes to RPO’s ethical behaviours, which, in turn, can attract talent, customers, and collaborators who value transparency and fairness.

Next, we conceptually apply these discussions to RPOs in the institute sector as an illustrative example in order to gain insights into how these RPOs can move toward an integrated strategy to research data governance.