Towards participatory multi-modeling for policy support across domains and scales: a systematic procedure for integral multi-model design

TL;DR

The paper tackles the challenge of policymaking for complex, cross-domain problems by proposing an integral MMS approach that combines multiple sub-models across scales and paradigms. It defines a systematic procedure to derive domain-knowledge requirements from multiscale and multi-paradigm literatures, and demonstrates the method with a Netherlands case study on school closures to reveal short-term and long-term cross-domain implications in healthcare and education. The procedure yields a transparent co-design process, leveraging CLDs and an MMS designed with a clear topology, coupling, and information exchange, enabling participatory involvement of domain experts. Findings indicate that MMS-based policy exploration can uncover trade-offs that single-domain or single-scale models would miss, supporting more informed and holistic policy formulation and design across domains and governance levels.

Abstract

Policymaking for complex challenges such as pandemics necessitates the consideration of intricate implications across multiple domains and scales. Computational models can support policymaking, but a single model is often insufficient for such multidomain and scale challenges. Multi-models comprising several interacting computational models at different scales or relying on different modeling paradigms offer a potential solution. Such multi-models can be assembled from existing computational models (i.e., integrated modeling) or be designed conceptually as a whole before their computational implementation (i.e., integral modeling). Integral modeling is particularly valuable for novel policy problems, such as those faced in the early stages of a pandemic, where relevant models may be unavailable or lack standard documentation. Designing such multi-models through an integral approach is, however, a complex task requiring the collaboration of modelers and experts from various domains. In this collaborative effort, modelers must precisely define the domain knowledge needed from experts and establish a systematic procedure for translating such knowledge into a multi-model. Yet, these requirements and systematic procedures are currently lacking for multi-models that are both multiscale and multi-paradigm. We address this challenge by introducing a procedure for developing multi-models with an integral approach based on clearly defined domain knowledge requirements derived from literature. We illustrate this procedure using the case of school closure policies in the Netherlands during the COVID-19 pandemic, revealing their potential implications in the short and long term and across the healthcare and educational domains. The requirements and procedure provided in this article advance the application of integral multi-modeling for policy support in multiscale and multidomain contexts.

Figures (8)

• Figure 1: Procedure for developing Multi-Model Structures (MMSs). First, develop a Causal Loop Diagram (CLD) that captures the relevant factors and relationships for the considered system and policy problem. Then, given the scale intervals associated with the factors, functionally decompose the CLD in sub-models at different scales (e.g., in time and governance level). Next, functionally decompose the sub-models by assigning adequate modeling paradigms informed by additional domain knowledge on, e.g., heterogeneity and discreteness in the system’s behavior. Finally, given the CLD and the functional decomposition in scale and paradigms, design an MMS consisting of the submodels and their interactions (topology of interactions, operations, and coupling templates).
• Figure 2: Conceptual Modeling: the Causal Loop Diagram (CLD) captures the implications of school closures for healthcare and education through a series of factors and relationships among them. Each relationship is assigned a number from 1 to 21 to facilitate their discussion. For healthcare, the core of this CLD is a Susceptible Infected and Recovered (SIR) model (1 - 10) kermack1927contribution with the addition of immunity loss (11) andeweg2022protection. School closures reduce contacts at school (12) among the school-age population (13), which contributes to the general number of contacts (14). For education, the CLD includes the negative effect of school closures on the cumulated learning of the school-age population (15) engzell2021learning. Such an effect is reduced by the percentage of the children’s families with a high income (16) engzell2021learning. Further, higher cumulated learning increases the percentage of highly skilled pupils (17) who have higher chances of obtaining high-income jobs (19 & 21), resulting in a higher percentage of high-income families oecd_education_2022. The percentage of high-income families improves the education opportunities of children from such families, thus increasing the percentage of highly skilled pupils (20).
• Figure 3: Scale and Paradigm Decomposition: The Causal Loop Diagram (CLD) obtained through Conceptual Modeling (factors and causal relationships among them) is decomposed into sub-models at different temporal scales (short- and long-term) and relying on different modeling paradigms (AgentBased Modeling (ABM) and System Dynamics Modeling (SDM)).
• Figure 4: Multi-Modelling Structure (MMS): the MMS (left) captured through the Multiscale Modeling Language or MML (right). The MMS shows the way the identified short- and long-term sub-models (cf. Figure \ref{['fig:scale-and-paradigm-separated-cld']}) exchange information and how such information is processed through operations (marked by mappers). For the MML, we refer the reader to Figure \ref{['fig:mml']} in the Appendix.
• Figure 5: Short- (left) and Long-term (middle and right) implications of school closure policies in the Netherlands. Three policies are considered. ”None” means that the school remains open at all times. ”Reff>1” means that schools are closed when the effective reproduction number is above one (assuming perfect information and no delays). ”Reff>1 & I>10%” is similar to the above, with the additional condition that at least 10 percent of the population becomes infected before closing schools. In the short term, policy effectiveness is measured in terms of the attack rate, expressed as the average number of times a person was infected during the pandemic. In the long term, policy implications are indicated by the “% of highly skilled pupils” (individuals younger than 20 years old that have a higher chance of finding highly paid jobs) and by the “% High Income” (percentage of the population older than 20 and younger than 50 that have a high income). For this illustrative example, it is assumed that no other exogenous factors (e.g., population or availability of highly skilled jobs) change over time throughout the simulation.