Adaptive interactive profit expectations using small world networks and runtime weighted model averaging

Paul William Bell

doi:10.1117/12.813941

Translator Disclaimer

30 December 2008 Adaptive interactive profit expectations using small world networks and runtime weighted model averaging

Paul William Bell

Author Affiliations +

Proceedings Volume 7270, Biomedical Applications of Micro- and Nanoengineering IV and Complex Systems; 727011 (2008) https://doi.org/10.1117/12.813941
Event: SPIE Smart Materials, Nano- and Micro-Smart Systems, 2008, Melbourne, Australia

Abstract

The aim of this paper is to simulate profit expectations as an emergent property using an agent based model. The paper builds upon adaptive expectations, interactive expectations and small world networks, combining them into a single adaptive interactive profit expectations model (AIE). Understanding the diffusion of interactive expectations is aided by using a network to simulate the flow of information between firms. The AIE model is tested against a profit expectations survey. The paper introduces "runtime weighted model averaging" and the "pressure to change profit expectations index" (p^x). Runtime weighted model averaging combines the Bayesian Information Criteria and Kolmogorov's Complexity to enhance the prediction performance of models with varying complexity but a fixed number of parameters. The p^x is a subjective measure representing decision making in the face of uncertainty. The paper benchmarks the AIE model against the rational expectations hypothesis, finding the firms may have adequate memory although the interactive component of AIE model needs improvement. Additionally the paper investigates the efficacy of a tuneable network and equilibrium averaging. The tuneable network produces widely spaced multiple equilibria and runtime weighted model averaging improves prediction but there are issues with calibration.

Citation Download Citation

Paul William Bell "Adaptive interactive profit expectations using small world networks and runtime weighted model averaging", Proc. SPIE 7270, Biomedical Applications of Micro- and Nanoengineering IV and Complex Systems, 727011 (30 December 2008); https://doi.org/10.1117/12.813941

ACCESS THE FULL ARTICLE