5th July 2017

Experience over the last decade shows that economies are prone to sudden contractions and then may grow more slowly than expected. This is the clearest example to suggest that economies may not be stable (at least in a local sense). Yet macroeconomic models are usually constructed such that optimal decision making by private agents guide the economy back to its stable growth path. This Discovery Meeting looks at changes that may be necessary, which avenues of enquiry might be most promising and why they may be resisted.

Different modelling approaches

In the mainstream model, short term fluctuations in output growth, or business cycles, arise because a representative agent optimally responds to unforeseeable technology shocks. In doing so the economy returns to its stable equilibrium growth path. To the extent that prices and wages may be slow to adjust, this may create a role for monetary policy. Government is limited to removing any rigidities and otherwise doing no harm. This is a marked departure from the economics of Keynes (1936) and earlier descriptions and models of internal economic cycles where booms sowed the seeds of subsequent busts (e.g. Minsky (1986) and Kaldor (1940)).

Alternative models with internal or endogenous business cycles exhibiting some of the persistent outcomes we observe in the real world have been developed. For example, a series of representative agent models by Farmer (1994, 2006) feature many possible outcomes, some of which are consistent with the concerns of earlier economists. Other non-linear models can also describe endogenous business cycles and persistent instability. The differences in these models can have profound effects on macroeconomic policy recommendations. The notion that the Phillips Curve may not be vertical after all opens-up questions about inflation targeting and can justify the use of fiscal policy. An important question is why there has been less acceptance of alternative approaches, even when the mainstream model appears to be contradicted by empirical evidence. Are there more promising lines of enquiry?

New micro-foundations

A defining feature of modern macroeconomic models is that they are grounded in optimal decision making by representative agents to ensure that the model is internally consistent. However, some economists question whether the specific micro-foundations are appropriate and are even required at all. When thinking about consumer preferences, it might be convenient to abstract from the influence of other agents, but this assumption may be inappropriate in macroeconomics. For example, the assumption that consumers make their own optimal decisions for all future periods is a far stretch from how psychologists and sociologists understand decisions are taken. Even taking biases and framing into account is at odds with how other disciplines understand decision making. For example, Akerlof’s (2007) suggestion of including norms in preferences maintains the purposefulness of micro-founded behaviour but re-establishes connections we observe in the real world such as wealth effects and the link between corporate investment and cash flows.

Other economists are even weary of the necessity of being bound to micro-foundations. Wren-Lewis (2012) argues that this prevents the use of empirical aggregate level relationships from being included in macroeconomic models. Clearly these relationships derive from our behaviour, but whether they are based on independent or even well understood decision making is another matter. Others argue for the inclusion of many different agents with different endowments and preferences and even more complex decision making. This introduces new methodologies such as agent based modelling techniques, but even here the decision making will require consistency. In early models, exchange between agents need not lead to a socially optimal outcome and thus justify government intervention. Is an alternative approach to the traditional conception of micro-foundations required?

Credit or financial cycles      

Records of credit cycles are even older than business cycles, going back to the early eighteenth century. Credit cycles are usually longer than business cycles – perhaps twice the length – but they have greater consequences when they bust in terms of economic cost (Taylor (2015)). The frequency of credit crashes or systemic financial crises has increased dramatically since the end of Bretton Woods system. Taylor notes that, for the seventeen leading economies, there were zero crises during the Bretton Woods era but 25 systemic crises since it ended. At the end of the last century the IMF found the spread of the Asia Financial Crisis to Russia and then Brazil as incomprehensible.

Mainstream macroeconomic models failed to include a meaningful financial sector. New models include credit frictions which can amplify temporary deviations. However, this is not the same as where financial sector decisions influence the allocation of resources. This would involve networked institutions (perhaps not deposit takers) that can increase their balance sheets and drive real economic activity. The very notion of financial instability and the costly dislocations from default have shown that there are long term consequences which are inconsistent with the assumption of stability. Governance arrangements around international capital flows remain relatively unexamined. Again, there is a history of accounts of credit cycles (e.g. Minksy (1986), Kindleberger (1978)) and a challenge is whether they can be included into core macroeconomic models.

Secular stagnation and balance sheets

Another set of ideas is around the concept of secular stagnation first raised by Hansen (1938) and revived by Summers (2013). The idea is that low levels of investment may be keeping output low and even reducing growth rates. This is consistent with the low rates of return and levels of investment. Excess savings that are not used for real investment result in low real interest rates which have consequences for the allocation of capital both into asset prices and across borders. The outcomes have potential consequences for financial and economic stability. In this model it is unlikely that monetary policy alone can deliver growth and financial stability.

This brings us back to why the low levels of investment persist at a time of great innovation. Some argue this depends on levels of debt from earlier expansions creating ‘balance sheet’ effects. Others suggest that the returns to those who make technological advances are so great that they are not recycled into effective demand. Earlier stock-flow consistent macroeconomic models may provide new insights. It is also worth noting that in these real world circumstances the mainstream models are surprisingly fragile (see Kocherlakota 2016).

17 July 2017