ReMoTe-S. Residential Mobility of Tenants in Switzerland: an agent-based model

Accounting for a fourth of the CO2 emissions and total energy consumption in Switzerland (IEA, 2018), housing plays a crucial role in the transition towards the sustainability of urban systems. Although new technologies have been developed to tackle the environmental burden of the residential sector, technical solutions alone are not sufficient to achieve the transition, as their success depends on how they can be reconciled with the present and future needs of households. Residential mobility offers the perfect ground to understand the matching of and recursive effect between households needs and dwellings. A wide range of models simulating residential dynamics exists; however, the specificities of the Swiss rental market require to consider factors influencing the decision to move and the selection of the new dwelling that are tenure- and context-specific, and which have not been modeled so far. In our presentation, we introduce ReMoTe-S, an agent-based model of the residential mobility of tenants in Switzerland. The model results from the interdisciplinary scientific collaboration of social, environmental and computer scientists. Based on exploratory qualitative and quantitative research, ReMoTe-S accounts for the complexity of tenants’ and buildings’ life course. The former includes changes in household (e.g. a new child), housing (e.g. an offer on the market), and work trajectories (e.g. a promotion), the corresponding triggers to move, and the consequent adjustment of residential preferences. The latter considers the construction, renovation and demolition of dwellings and buildings of two types of Swiss multifamily housing providers: cooperatives and asset managers. The matching between household- and dwelling-agents is determined by preferences and constraints. Preferences are simulated as desired housing functions (e.g. a shelter, a status symbol); their correspondence with the function fulfilled by the dwelling is used to determine households’ satisfaction, and thus the relocation behavior. While the desired function is specific to each household type (e.g., ‘young single’), the dwelling function is computed from the dwelling, building and neighborhood characteristics (e.g., ‘proximity to public transports’). Constraints are set based on rules specific to the Swiss rental market (e.g., occupancy rules, salary-rent proportion). The model simulates an open market, where in- and out-migration are responsible for the variation in dwellings’ vacancy rate. To check for plausibility and robustness of the baseline scenario, calibrate the model and run experiments, the effect of the variation of several parameters were explored. Our presentation shows the results of the OFAT or One-Factor-at-a-time sensitivity analysis, where, to account for the stochastic variation of parameter values and estimate long term behaviours independent of initial patterns, each simulation run was repeated 100 times over a time span of 30 years (or 360 time-steps). In this context, two what if experimentations are briefly presented, aimed at investigating the reciprocal effect of household preferences on dwellings in the framework of housing sustainability. The first explores how variations in the size of shared flats influence individual space consumption in unpredictable ways; the second looks at changes in vacancy rate resulting from changes in the characteristics of new dwellings on the market. The final goal of the presentation is to discuss with the audience the last crucial step before the model is available to be shared, integrated and used: the validation. This phase is strongly debated in the literature on ABM, which offers a large amount of possibilities. The introduction of the model, its verification, and the two what if experimentations will be used as a base to discuss the following questions: Against what would this empirical agent-based model be validated? Which approach would be the best to validate it? Should the validation occur now, or in a later stage? The last question refers to the larger scope of the model. More precisely, ReMoTe-S was conceived as one piece of a larger interdisciplinary collaboration, where the ABM of households’ residential mobility would be integrated with a second one, simulating more accurately owners’ strategies (i.e. investment, demolition), and a third one, calculating the resulting environmental footprint. The integration of these three models would allow for the exploration of scenarios that take into account, simultaneously, the socio-cultural, economic and environmental aspects of housing’s sustainability