The ﬁrst part of the paper is a brief introduction to the concepts and methods used in recent endogenous business cycles models. Endogenous deterministic and stochastic fluctuations are bound to occur, under increasingly plausible assumptions, in models with individual optimization, market clearing and self-fulﬁlling expectations when there are capital market imperfections. The phenomenon is most likely to be observed, in a nonlinear framework, when some eigenvalue(s) of the system have a modulus close to 1 (unit roots). It is argued that endogenous business cycles models have become more and more credible alternatives to describe observed fluctuations in our economies. The second part of the paper reviews recent studies suggesting that self-fulﬁlling expectations are often dynamically unstable when learning is taken into account. The phenomenon is most likely to occur when expectations matter signiﬁcantly, which might explain why actual economic time series display higher volatilities in markets for capital investment, inventories, durable goods, ﬁnancial assets and stocks. It is suggested that on account of the important nonlinearities involved in learning, actual learning dynamics may generate highly complex, even chaotic, trajectories.
Grandmont, Jean-Michel, "Expectations Driven Nonlinear Business Cycles" (1992). Cowles Foundation Discussion Papers. 1265.