eduzhai > Applied Sciences > Computer Science >

A Comparison of Economic Agent-Based Model Calibration Methods

  • KanKan
  • (0) Download
  • 20210425
  • Save

... pages left unread,continue reading

Document pages: 32 pages

Abstract: Interest in agent-based models of financial markets and the wider economy hasincreased consistently over the last few decades, in no small part due to theirability to reproduce a number of empirically-observed stylised facts that arenot easily recovered by more traditional modelling approaches. Nevertheless,the agent-based modelling paradigm faces mounting criticism, focusedparticularly on the rigour of current validation and calibration practices,most of which remain qualitative and stylised fact-driven. While the literatureon quantitative and data-driven approaches has seen significant expansion inrecent years, most studies have focused on the introduction of new calibrationmethods that are neither benchmarked against existing alternatives norrigorously tested in terms of the quality of the estimates they produce. Wetherefore compare a number of prominent ABM calibration methods, bothestablished and novel, through a series of computational experiments in anattempt to determine the respective strengths and weaknesses of each approachand the overall quality of the resultant parameter estimates. We find thatBayesian estimation, though less popular in the literature, consistentlyoutperforms frequentist, objective function-based approaches and results inreasonable parameter estimates in many contexts. Despite this, we also findthat agent-based model calibration techniques require further development inorder to definitively calibrate large-scale models. We therefore makesuggestions for future research.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×