DerivSIM- A Cross-Impact Based Stochastic Simulation Method for Forecasting and Policy Analysis
DerivSIM: A Cross-Impact Based Stochastic Simulation Method for Forecasting and Policy Analysis
In any given context, policy makers may be concerned with forecasting how a set of important variables will evolve into the future. Following the general spirit of cross-impact analysis, DerivSIM, the method proposed in this paper describes that evolution in terms of cross-impacts, which are mutual interactions and relationships between pairs of variables.
DerivSIM, short for “derivative simulation”, essentially regards each cross-impact pair as a mathematical derivative which specifies how a change in one variable directly impacts another. However, unlike most of the literature in which cross-impacts are described by deterministic numbers, our method introduces an innovative random element to the cross-impact specifications which renders greater realism. Also, our approach differs from several others in taking a (stochastic) structural dynamics approach which necessarily models time. The convenience of such framework enables us to allow for delayed cross-impact effects.
Finally, our framework provides a simple way to model different policies as exogenous shocks to the system of variables, and analyze their impact during the time horizon of interest.