Drawing of an looking through binoculars at graphs
VoxEU Column Macroeconomic policy

Dynamic disequilibrium with randomness

A key objective of modern macroeconomics is to understand, characterise, and provide policy guidance for economic crises. Nevertheless, the standard dynamic stochastic general equilibrium models struggle to account for deep and persistent downturns, such as the 2008 crisis. This column presents a dynamic disequilibrium theory with randomness to provide new insights into these episodes. The framework features incomplete markets, individual heterogeneity, and an evolving economy in which people are constantly learning. In addition to being more realistic, the theory enables the design of policies that will mitigate the frequency of crises and ameliorate their consequences.

The central problem of macroeconomics is to explain the deep, and often persistent, downturns accompanied by high levels of unemployment that episodically afflict capitalist economies. These phenomena motivated the emergence of the field after the Great Depression. Macroeconomic crises are extreme examples of economic fluctuations and are the most relevant to people’s lives. They are also the events that teach the most about key properties of the economic system, in a way that small inventory cycles do not.

Some crises are simply the result of external shocks, for instance, a natural disaster that destroys a large part of the country’s capital. But most fluctuations are not the result of such an exogenous shock or of a random shock to productivity, as characterised by dynamic stochastic general equilibrium (DSGE) models, whether of the New Keynesian (e.g. Christiano et al. 2005) or real business cycle (RBC) (Kydland and Prescott 1982) variety. Rather, most fluctuations are associated with endogenous ‘shocks’, including large and often sudden changes in views about the workings of the economy. The US Great Recession in 2008, the crises of Argentina in 2001 and 2018, and the Greek crisis from 2008 onwards are all examples. In those cases, it is not possible to identify an exogenous technology shock that tripped the economy from prosperity into a deep downturn. Rather the events, characterised by bankruptcies and defaults as well as by high and persistent unemployment, show a malfunctioning of the economic system. Not surprisingly, DSGE models have failed to predict even the possibility of those downturns, to explain them, or to design appropriate policy responses. 1

We are forced, then, to identify which of the many unrealistic assumptions in those models are most central to their inability to provide a good account of macroeconomic fluctuations. Is it rational expectations? The assumption of a representative agent? The absence of asymmetric information? That the most important shocks giving rise to fluctuations are exogenous technology shocks rather than shocks created by the market itself?

In explaining deep downturns, we need to understand (a) how the market economy generates such large fluctuations in aggregate demand disproportionate to any exogenous shocks in the ‘real variables’ and the ability of labour markets to adjust, thereby giving rise to high unemployment and (b) the dynamics of adjustment: why they are such that high levels of unemployment can persist. In Guzman and Stiglitz (2021), we present a dynamic disequilibrium theory with randomness that provides insights into the underlying economic processes in ways that the DSGE models cannot, and in doing so provides policy guidance both on how to make deep downturns less frequent ex-ante and less severe ex-post.

Situating the problem in a historical context

By construction, macroeconomic instability and disequilibrium are absent from the standard benchmark framework of neoclassical economic theory—Arrow-Debreu. That framework relies on the assumption of the existence of a complete set of markets, which, in turn, implies that nothing constitutes a perturbation and that there are never situations in which the set of plans turns out to be inconsistent with the set of budget constraints of the market participants—implying, in turn, that there are never defaults or bankruptcies.

The limitations of market mechanisms in coordinating intertemporally the actions and decisions of the interacting economic decisionmakers onto a stable path of aggregate demand with full employment was, of course, a central concern of macroeconomics in the years that followed the Great Depression. John Maynard Keynes wrote before Kenneth Arrow and Gérard Debreu’s fundamental work, so he did not have the powerful apparatus they provided that has been so extensively used by mainstream economists since the mid-1950s to pin down what deviations from the perfect markets’ benchmark could lead to the observed frequent and persistent macroeconomic dysfunctions. The mainstream macroeconomics literature that followed from the late 1970s on, by assuming that the system is always in a state of macro-consistency, simply assumed away the most important questions of the field. That literature looked to frictions like price rigidities as the culprit for the malfunctioning of the economic system. But this oversimplified the sources of macroeconomic fluctuations and provided poor policy guidance, suggesting that if only we eliminated these price rigidities all would be well. 

First central problem: Macroeconomic inconsistencies and disequilibrium

While dynamic stochastic general equilibrium models do not assume a complete set of Arrow-Debreu securities, they assume that transversality conditions hold in every possible state, implying a similar result. The first central failure, we contend, is that they provide no explanation for why that assumption should hold in a decentralised economy. With incomplete markets it is not possible to check that all plans will be consistent in every state (i.e. that there would be market clearing in those markets, were they to have existed), and it is likely that they will be revealed not to be consistent in some state at some date.

Of course, in a representative agent model there is no problem in ascertaining the consistency of plans—there is no need for markets (and if the model were accurate, there would be no markets) and therefore no consequence for the absence of markets. But when agents differ in a decentralised economy, there is, in general, no way for them to know other agents’ plans for the future and therefore no way for them to know—in the absence of markets extending infinitely far into the future—that they are consistent. Moreover, even if there were common knowledge about agents’ current plans, in an evolutionary non-stationary environment those plans would be incomplete. So, it would not be assured that they are consistent in the future.

The dynamic stochastic general equilibrium models assume that the moment after the economy experiences an unanticipated shock, like in 2008, it magically moves to a new rational expectations equilibrium trajectory—and, with marked cognitive dissonance, under the belief that such an unexpected shock will never occur again. Whether there is a stationary world in which the assumption that the economy is always on an equilibrium trajectory (except when it discovers it is not) might make sense is not our concern: We live in a non-stationary world that is constantly evolving, and it is clear that in such a world the assumption of ever-present equilibrium does not make sense.

Second central problem: Disequilibrating dynamics

Even with a full set of markets, neoclassical theory was never able to establish under general conditions the stability of the market equilibrium, i.e. plausible dynamics such that when the economy is perturbed from its equilibrium it would return to equilibrium ever, let alone quickly. We analyse natural decentralised dynamics, showing that, to the contrary, under plausible conditions they may be disequilibrating in the short run. For instance, wage and price adjustments might exacerbate the inconsistencies that are revealed after a shock, through increases in the real values of debt or decreases in aggregate demand, and thus be ineffective in restoring market equilibrium.

Third central problem: Endogenous shocks and non-stationary stochastic processes

We argue further that the central problem in understanding large episodic unemployment is understanding the nature and magnitude of the shocks to aggregate demand, rather than wage rigidities. Partly, this requires understanding how financial frictions and other market rigidities amplify shocks (as emphasised in Greenwald and Stiglitz (1993), for example). In addition, it requires understanding how heterogeneous beliefs and changes in the degree of dispersion give rise to large fluctuations in aggregate demand (as in Fostel and Geanakoplos (2008) and Geanokoplos (2010) in their theory of leverage cycles, and Guzman and Stiglitz (2021) in their theory of pseudo-wealth). And lastly, it requires a better understanding of how economic agents’ expectations respond to an ever-evolving economy, making inferences as new data become available, and, especially, responding to macroeconomic inconsistencies as they get exposed (e.g. Leijonhufvud 1981).

Essential ingredients of a dynamic disequilibrium theory with randomness

Thus, the essential ingredients in our analysis are (i) an incomplete set of markets, (ii) individual heterogeneity, and (iii) an evolving economy in which individuals are constantly learning about the structure of the economy and the behaviour of agents within it. This learning, of course, affects expectations formation and behaviour, leading, in turn, to further evolution of the economic system.

In the absence of economic evolution, itself partially based on endogenous changes in technology, we would presumably eventually learn fully about the economy, but because the economy is always evolving there is always learning; and because of the persistence of incomplete information and information asymmetries—different individuals see and perceive different information and process it differently, so there is persistence in differences in beliefs—we never attain the utopia of common knowledge. Individuals are exposed to different signals and process the information in different ways. And large events that reveal that pre-existing assumptions about the economy are not valid give rise to large changes in beliefs. In turn, those changes in beliefs, as well as the unanticipated and non-anticipatable real changes that give rise to them, lead to situations in which the full set of contracts cannot be honoured, to such an extent that there is, in some cases, a crisis.


Policy guidance needs an analysis of real dynamics, including how to stabilise aggregate demand (through better-designed automatic stabilisers and restricting destabilising speculation) and how to quickly restore the economy to full employment when macroeconomic inconsistencies get revealed (for example, through better debt restructuring mechanisms). When it otherwise becomes apparent that widespread beliefs about the economy may not be valid, giving rise to heightened uncertainty and concomitant precautionary behaviour, the adverse macroeconomic effects can be offset through the design of better public risk-sharing mechanisms. 

Dynamic disequilibrium theory with randomness provides an alternative framework for macroeconomic analysis and research, more likely than prevalent dynamic stochastic general equilibrium models to enhance our understanding of the major fluctuations repeatedly afflicting the economy, and more likely to enable the design of policies that will mitigate their frequency and ameliorate their consequences.  


Christiano, L J, M Eichenbaum and C L Evans (2005), "Nominal rigidities and the dynamic effects of a shock to monetary policy", Journal of Political Economy 113(1): 1-45.

Colander, D, P Howitt, A Kirman, A Leijonhufvud and P Mehrling (2008), "Beyond DSGE models: toward an empirically based macroeconomics", American Economic Review 98(2): 236-40.

Fostel, A and J Geanakoplos (2008), "Leverage cycles and the anxious economy", American Economic Review 98(4): 1211-44.

Geanakoplos, J (2010), "The leverage cycle", NBER macroeconomics annual 24(1): 1-66.

Greenwald, B C and J E Stiglitz (1993), "Financial market imperfections and business cycles", The Quarterly Journal of Economics 108(1): 77-114.

Guzman, M and J E Stiglitz (2020), "Towards a dynamic disequilibrium theory with randomness", Oxford Review of Economic Policy 36(3): 621-674.

Guzman, M and J E Stiglitz (2021), "Pseudo-wealth and consumption fluctuations", The Economic Journal 131(633): 372-391.

Korinek, A (2018), "Thoughts on DSGE Macroeconomics: Matching the Moment, But Missing the Point?", in Toward a Just Society, pp. 159-173, Columbia University Press.

Kydland, F E and E C Prescott (1982), "Time to build and aggregate fluctuations", Econometrica 50(6): 1345-1370.

Leijonhufvud, A (1981), Information and coordination: essays in macroeconomic theory, Oxford University Press.


  1. See for instance the papers of Oxford Review of Economic Policy, Volume 34, Issue 1-2, Spring-Summer 2018, Rebuilding macroeconomic theory and Volume 36, Issue 3, Autumn 2020, Rebuilding macroeconomic theory II, as well as Colander et al. (2008) and Korinek (2018).

2,055 Reads