Bence Mérő - Size Does Matter – The Optimal Choice of Scaling in Economic Agent-Based Models

  • Presenting author: Bence Mérő (Central Bank of Hungary, Complexity Science Hub Vienna)

  • Authors: András Borsos, Zsuzsanna Hosszú, Bence Mérő, Nikolett Vágó, Zsolt Oláh

  • Session: B01C - Agent-Based Modelling - Tuesday 9:00-10:30 - Senate Hall

  • Slides: PDF

There are two typical strategies in economics to reduce the size and complexity of the models: (i) using representative agents by aggregating the actual entities, (ii) and downscaling, i.e. using only a sample of agents. While the first strategy has been studied in detail in mainstream economics, the implications of the second option – which is mainly used in complexity economics – are underresearched. This paper contributes to filling this gap in a twofold way. First, we identify two main channels via which scaling can influence complex economic agent-based models: (i) idiosyncratic shocks and (ii) information loss due to insufficient interactions. Second, we analyse the implications of these mechanisms through assessing the trade-offs between three fundamental measures of model performance: precision, accuracy and running time with different downscaling levels ranging between 0.25% - 100% of the full population. To conduct this analysis we use the model of Mérő et al. (2022), which is suitable to represent the housing market of Hungary at any scale in this interval (from 10,000 to 4 million agents). We show that there is a non-trivial relationship between the scaling factor, the model performance and the running time. Not only the model’s accuracy and precision depends on the model size in a non-linear way, but we also found that the evaluation of a scenario at a given level of precision takes only 3-4 times longer with 100 times more agents. Based on our results we claim that the optimal choice of model size could be greater than what is commonly used in the literature.