Technocratic Decision-Making in Times of Crisis? The Use of Data for Scientific Policy Advice in Germany’s COVID-19 Management

Regarding the above-mentioned conflicting rationalities, scholars distinguished the “two worlds” of scientific policy advice on the one hand and politico-administrative decision-making on the other (Habermas, 1969).⁴ The academic debate evolved around the peculiarities, specific logics, yet also increasingly the interface, boundaries and boundary-spans between these worlds. In the so-called “decisionist model”, politico-administrative actors are conceived as dominating the process and making use of experts’ knowledge in a purely instrumental way, assuming that this knowledge is value-free. The “technocratic model”, by contrast, assumes a superiority of scientific expertise based on the belief that science can identify optimal solutions to every problem, thus reducing politico-administrative actors to mere executors of scientific directives (Böcher, 2007; Martinsen & Rehfeld, 2006; Schelsky, 1979; Weber, 1988). These ideal types, at best partially representing reality, have been contrasted with the so-called “pragmatic model” as a third, more realistic model. Accordingly, politico-administrative decisions are seen as results of an iterative process between actors actively involved either in scientific policy advice or in decision-making. This implies an ongoing “translation process between science and politics” (Habermas, 1969: 137). Metaphors such as “boundary spanning” (see e. g. Martinsen & Rehfeld, 2006: 48) have been used to describe this particular process. More recent concepts, such as the recursive model of scientific policy advice (Weingart, 2001) or the concept of “boundary work” (Korinek & Veit, 2013), similarly emphasise the complex interdependent relationships and negotiation processes between science and politics.

The concept of “boundary work” builds on assumptions such as the above-mentioned importance of knowledge as a resource of bureaucratic power postulated by Weber (2006: 226). Also linked to Habermas’ pragmatic model of scientific policy advice, the concept assumes that both politics and science aspire for expansion into each other’s area of competence. Consequently, the respective boundaries between science and politics are subject to constant negotiations, thus resulting in scientific policy advice being an essential part of a “boundary world” (Gieryn, 1983: 791f; Korinek & Veit, 2013: 267, Sokolovska et al., 2019). As a result of this, scientific policy advice is to be characterised by a high degree of uncertainty of respective expectations. This structural feature of scientific policy advice can significantly complicate data-driven policies (Gerlinger, 2019; Haucap, 2020; Weingart, 2019). Therefore, the involvement of ‘brokers’ who provide the necessary ‘translation services’ is required.

Illustrating the need for such brokers, it is to be stressed that scientific policy advice often relies on contextualization and nuanced interpretations, whereas policy-makers expect unambiguous interpretations, clear conclusions as well as recommendations for practical implementation or even implement trial-and-error strategies (Boin & Lodge, 2021). Further complication arises from the fact that policy-makers tend to normatively contextualize scientific data in order to allow for necessary considerations with other normative criteria based on societal expectations (Gerlinger, 2019). Hence, “the same data (can) also lead to several translations of different sense”, so that the “ambiguity of data (…) opens up the possibility of different interpretations” (Schwab, 2014). Correspondingly, the need for such negotiation processes in the context of data utilization “challenges the rational decision-making model of performance-oriented reform concepts” (ibid.). Various authors (e.g., Moynihan, 2008, cited in Schwab, 2014) therefore propose procedures of discursive and interpretive processing of data or refer to the context-dependency of data utilization.

To explain data utilization in crisis-related advisory and decision-making processes, institutional contexts must be taken into account. It is assumed that varying degrees and scopes of data utilization in organizations are shaped by formal organizational features and regulations on the one hand and administrative cultures and informal rules on the other (see Christensen & Lægreid, 2020; Kuhlmann et al., 2021). Whereas formal regulations (e.g., privacy rules) and structures (e.g. organizational hierarchies, digital maturity of processes) shape the macro-setting of data-based advice and limit corridors of action, informal norms (e.g., data culture, data literacy) and identities generate a “logic of appropriateness” (March & Olsen, 1989) which impacts on the meso- and micro-levels of organizations. Against this background, e-government research has identified process-related, technological as well as legal hurdles of data utilization (Kempeneer, 2021; Matheus et al., 2020). Different types of data strategies (e.g., internal and external orientation, cf. Van Donge et al., 2020) may determine data utilization, as they institutionalize the data culture of the respective institutions. In this context, the digital maturity of public organizations (see Kuhlmann & Heuberger, 2021) represents a key variable to explain data utilization in advisory settings. The varying levels of progress in the digital transformation of public administration thus appear to be relevant explanatory factors for the extent and intensity of ICT-based data utilisation in crisis-related policy advice. Of course, the relationship between data utilization on the one hand and institutional contexts on the other hand also exists vice-versa: institutions may not only shape data utilization, as the latter might also shape the respective institution (Christensen & Laegreid, 2020: 4).

In our analytical framework, the interface models (technocratic, decisionist, pragmatic) are conceived as dependent variables of investigation, while the context conditions (data management, digitalization, data culture) are considered as explanatory variables. We assume that the institutional context influences the mode of policy advice and decision-making. More precisely, favourable conditions for data-based scientific policy advice are expected to promote technocratic policy-making, as investments in such context conditions suggest a demand for data-driven and therefore rather technocratic policy-making. This refers inter alia to Germany’s comprehensive efforts towards the digitalization of public administration and the accompanying financial support to speed up the progress of digital transformation in the public (health) sector (see Kuhlmann & Heuberger, 2021). Furthermore, the public health service with its well-established and internationally renowned surveillance and reporting system, as well as Germany’s position as a frontrunner in health-related research (e.g., in virology, epidemiology) suggest a favourable context for data-based scientific advice and policy-making. Because of the particularities of the COVID-19 crisis, especially the importance of data as a strategic key resource in decision-making, we expect these conditions to facilitate a predominantly technocratic model of policy advice in which scientific advisors and data-providers are privileged vis-à-vis to policy-makers. Respective expectations are backed up by first insights from public administration literature in Germany, underlining the hypothesis of a predominantly technocratic mode of policy advice and policy-making in Germany (cf. Böcher et al., 2021). Going one step further, Florack et al. postulates a “Coronacracy” (Florack et al., 2021). In addition, the uncertainty and volatility of data and knowledge especially at the beginning of the crisis pushes the demand for scientific policy advice in order to legitimize political decisions in such contexts, thus paving the ground for a technocratic dominance in the policy-making process.

Irrespective of different rationalities, data-based scientific policy advice and decision-making are foremost dependent on the availability and accessibility of scientific data. However, the COVID-19 crisis has demonstrated that these requirements have not been met sufficiently. Today, there are still significant data gaps that impede evidence-based advice as well as decision-making. Despite of complements to existing surveillance and health monitoring systems initially established for non-crisis periods, considerable quality problems and data deficiencies have come to light.

The interplay between policy-makers and scientific data providers as well as the use of data for the preparation and legitimation of decision-making processes has never been more pronounced than during the COVID-19 crisis (Interview 5). Consequently, the operational rationality of policy-makers has been and continues to be highly “data- and number-driven” (ibid.). Data collection and analysis are not the only central components of crisis management, keeping in mind that the overarching policy goal is to protect the health care system from overload. Policy-makers also see the necessity to justify their decisions by referring to scientific data and evidence, which is considered as a decisive factor for the society’s acceptance of measures and their willingness to comply.

At the federal level, institutionalized contacts with the scientific data providers of the departmental research institutions and with the scientific advisory boards of the ministries facilitated the integration of data in the decision-making process. However, political decisions on measures were largely based on external data providers, especially on virological and epidemiological expertise, but also on physical modelling and forecasts. While this already refers to the criticism on an ‘expertocracy’, further problems arose due to the media presence of some external knowledge providers, thus determining the political decision-making process. Due to the effects of public expectation, this had an impact on the scope of political action, sometimes hindering deliberative, data-based decision-making (Interview 10).

In accordance with the postulated need for ‘brokers’ and ‘translation’ (see above), major communication problems and uncertainties in the mutual expectations of scientific data providers and political-administrative decision-makers were observed, particularly during the initial phase of the COVID-19 crisis. Mutual understanding between both ‘worlds’ turned out to be difficult, as the “same language was not spoken” (Interview 11). Furthermore, an uncertain and volatile data situation combined with a dynamic development of the crisis complicates the weighing of data in the context of decision-making processes. In this situation, scientific statements characterised by contextualization and deliberative wording contrasted the demand of political-administrative decision-makers seeking to obtain unambiguous findings and precise recommendations. While scientific data providers have to emphasise the incertitude of forecasts and to provide complex explanations on the data context in order to meet the criteria of scientific working, this results in uncertainty for data users in politics and public administration. Due to such communication problems between scientific data providers and political-administrative decision-makers, it remains unclear whether and to what extent scientific data ultimately influenced and substantially determined the decision-making processes. Nevertheless, a clear learning curve and a mutual convergence between scientific data providers and political-administrative decision-makers were observed in the course of the COVID-19 crisis.

However, there are still deficits in the mutual understanding of respective roles between scientific data providers and political-administrative decision-makers. At various points, urgent calls were made for science as well as for politics to reflect on their respective roles. Thus, science has “advantages concerning its knowledge, but no democratic legitimacy” (Interview 8). Consequently, scientists were claimed to advertise and to convince political-administrative decision-makers in order to push forward their proposed solutions. In contrast, the latter were expected not to demand ready-made decisions from scientists, as their function consists of balancing and weighing normative values, taking into account different data-based interpretations and potentially conflicting societal interests when choosing one policy option while rejecting another one.

As mentioned above, the validity of knowledge concerning the COVID-19 pandemic is characterised by an unprecedented short-term, volatile and contested nature. This resulted in further problems, especially for political-administrative decision-makers: “it has never happened before, that every 6-8 hours the scientific recommendations changed completely” (Interview 10). Consequently, containment measures were sometimes adjusted in parallel. The discussion on closures of schools illustrates respective challenges: initially, policy-makers refrained from this measure citing scientific policy advice. As the scientific debate evolved and political perceptions changed, school closures were finally enacted irrespective of their initial rejection. The uncertainty and volatility of data and resulting policy recommendations, manifested in short-term, sometimes even fundamental changes of the current state of the academic debate, are characteristic features of this particular crisis, profoundly challenging consistent data-based decision-making.

In addition, various interviewees criticized the fact that political decision-makers at the federal and Länder levels failed to ensure sufficient transparency on the data and evidence they took into account to formulate containment measures. Given that the particular potential of data consists of ensuring acceptance of political measures, a lack of transparency about data relevant to decision-making and justifications for measures characterised by an insufficient database appears to be a particularly serious omission. This is mostly criticised by local decision-makers, as the above-mentioned aspects resulted in confusion and a lack of comprehension for rapidly changing measures. Thus, respective actors face increasing difficulties in convincing the local population of the meaningfulness of the measures and the need to comply with them.

A further criticism refers to the lack of transparency regarding the selection of data providers and experts as well as their respective mandates (Interviews 2 and 6). As a result, the answer to the question of which data was perceived by political decision-making too often depended on “who is closest to the German federal minister of health Jens Spahn” (Interview 6). Against this background, there is a call for the selection of data providers and experts to be made with greater transparency and more aligned with scientific criteria (recommendations, peer reviews, publications, etc.). Another problem is seen in the limited informational capacities and increasing information overload of politico-administrative actors, who face severe difficulties in “digesting” and processing all crisis-relevant information inputs adequately. Scientific advisors (particularly coming from departmental research institutions) complain about the “clogging of the pipeline” for data providers because policy-makers could not process the ever-growing loads of data and thus some serious bottlenecks of data transfer became apparent (Interview 8).

Data utilization largely depends on a supporting or hindering institutional infrastructure and administrative culture (see above). “Digital maturity” (cf. Coursey & Norris, 2008) and the “data culture” of administrative organizations (e.g., federal ministries), as well as the “data literacy” of involved actors are key variables to explain data (non-)utilisation in crisis-related policy advice and decision-making. It turned out that data utilization is frequently hindered by a missing awareness of the data available in-house and/or of its strategic relevance for organizational processes (Interview 11). Yet, even in cases of more extensive awareness, there are still challenges of data sharing across organizational boundaries. The extent to which data-based scientific policy advice can actually reach policy-makers largely depends on the extent and quality of (digital) data sharing. In this context, significant deficits and bottlenecks had become evident even before the COVID-19 crisis. “Departmental silos” (see Bundesregierung, 2021: 59) restrain the access of various departments to shared databases and the cross-departmental re-use of data. Data tends to be ‘hoarded’ in one’s own organisation instead of being shared with other administrative units.

The extent and intensity of cross-organisational data exchange largely depends on the respective data tradition of the organisation and the professional backgrounds. Administrative departments with predominantly legal professional backgrounds and/or little experiences in data-driven processes have proven to be less cooperative and open to data-sharing. In contrast, those departments with a long-standing practical experience in dealing professionally with larger volumes of data on a daily basis and/or professional backgrounds rather dominated by data and social scientists are more open to data-sharing and less inclined to data “hoarding”.

Furthermore, the fostering of competencies regarding the legal context of data utilization and data privacy rules is necessary. Even in cases of a predominant data-based operational rationality, there is widespread uncertainty on the scope for action regarding the regulatory conditions of data utilization and data sharing because it is unclear what is legally permitted and what is prohibited. The certainty of action with regard to the question of “whether and how to use data” (Interview 11) and an administrative culture anchoring data-oriented policy-making are still extremely limited in public administrations. These application hurdles of data protection rules, rather than the regulations as such, are the main source of problems with data utilization in public administrations.

Moreover, there is a lack of machine-readable provision of raw data, which leads to deficiencies in demand-responsiveness in data management. Due to this limitation, it has only been possible to use data for its intended purpose, whereas the ‘re-use’ of data in different contexts and for different problems has only occurred to a limited extent. Furthermore, administrations are lacking the necessary institutional arrangements and organizational structures for data-driven processes, both for their standard operations and for crisis management. Consequently, important data compilations and checks are often carried out manually, which causes additional personnel expenses and does not make work easier, which also represents an “add on” to normal working procedures. Thus, the growing need in politics and administration, especially in times of crisis, to obtain data across levels and organizations and to feed it into decision-making processes cannot be satisfied due to the lack of functioning process structures.