A forecast built from interacting agents captures feedback loops — demand shifts affect firms, which affects employment, which affects demand again — that a single smooth curve can't.
output(t+1) = output(t)*(1+trend) + shock*volatility
- Household agents — consumer agents whose spending drives demand.
- Firm agents — producer agents whose output responds to demand and shocks.
- Market volatility — size of random shocks buffeting the modelled economy each step.
- Trend momentum — how strongly recent growth or contraction persists into the next period.
Central banks run agent-based models alongside traditional econometric ones specifically because interacting-agent dynamics reproduce boom-bust patterns that smooth equation models miss.