Abstract: Keynesian economists refer to capitalism as a monetary production economy, in which the theory of money and the theory of production are inseparable (Skidelsky, 1992). One important aspect of this, brought to light by Robertson following the publication of The General Theory, is that in a Keynesian economy, endogenous money creation is logically necessary if the economy is to expand. A Keynesian economy cannot operate with an exogenously given supply of money as in verticalism. One way to ensure that money is endogenous is to simply assume that the supply of money is infinitely elastic, known in the literature as horizontalism. In this view, prior savings cannot be a constraint on current investment and it follows that the level of economic activity is determined by effective demand. Using a multi-agent systems model, this paper shows that real economies, especially those subject to recurrent financial crises, can be neither horizontalist nor verticalist. Horizontalism overlooks microeconomic factors that might block flows from savers to investors, while verticalism ignores an irreducible ability of the system to generate endogenous money, even when the monetary authority does everything in its power to limit credit creation.
Abstract: Agent-based models are inherently microstructures–with their attention to agent behavior in a field context–
and only aggregate up to systems with recognizable macroeconomic characteristics. One might ask why
the traditional Keynes-Kalecki or structuralist (KKS) model would bear any relationship to the multi-agent
modeling approach. This paper shows how KKS models might benefit from agent-based microfoundations,
without sacrificing traditional macroeconomic themes, such as aggregate demand, animal spirits and endogenous
money. Above all, the integration of the two approaches gives rise to the possibility that a KKS
system–stable over many consecutive time periods–might lurch into an uncontrollable downturn, from which
a recovery would require outside intervention. As a by-product of the integration of these two popular
approaches, there emerges a cogent analysis of the network structure necessary to bind real and financial
agents into a integrated whole. It is seen, contrary to much of the existing literature, that a highly connected
financial system does not necessarily lead to more crashes of the integrated system.