Fiscal multipliers, expectations and learning in a macroeconomic agent‐based model

This paper evaluates the government expenditure multiplier and the influence of agents' expectations and consumption choices thereupon in a pre-existing estimated macroeconomic agent-based model. If the simple consumption heuristic of the baseline model is replaced by inter-temporal optimization subject to a budget constraint based on agents' estimations of future income, the multiplier becomes significantly smaller. When agents' beliefs about the effects of expenditure shocks are explicitly introduced, they can strongly increase or decrease the multiplier. If agents are allowed learn about the effects of government expenditure on their income from repeated shocks, they are able to correctly predict these effects.