Haldane outlines ‘interdisciplinary’ economics model

Andy Haldane and Arthur Turrell describe a way of broadening the scope of economics’ limited array of core models
People against city backdrop
Agent-based models seek to capture the diversity of people in an economy

The Bank of England’s chief economist has joined forces with a specialist in plasma physics to outline a way of opening up the “insular” discipline of economics.

Andy Haldane and Arthur Turrell, a physicist-turned-data scientist at the BoE, present agent-based modelling as an interdisciplinary alternative to the standard models of economics, in a staff working paper published today (November 24).

Economics scores poorly in terms of citing other disciplines, and economists suffer from lower levels of trust among the general public than natural scientists, Haldane and Turrell write. “Macroeconomics has much to gain from taking inspiration from other disciplines; and other disciplines could in turn benefit from a better understanding of economics,” they say.

They posit that the “macroeconomic mono-culture” emanated from the “new classical counter revolution” led by Robert Lucas, Edward Prescott and others in the 1970s. These economists argued that working with models grounded in “microfoundations” could overcome the failure of econometric models to cope with shifts in policy.

Later, microfoundations came to be rooted in a “particular kind of self-interested behaviour” – known as rational expectations – that came to underpin models central to macroeconomics, namely the dynamic stochastic general equilibrium (DSGE) model used in central bank policymaking and economics faculties across the globe.

Other disciplines have moved on to adopt a wider set of models, Haldane and Turrell say. For example, “modern physics research deals with complex systems, emergent behaviours, vast simulations and outcomes which are probabilistic and stochastic beyond what is implied by the Gaussian distribution”. Economics is only just starting to catch up.

‘The model from Monte Carlo’

The authors note there are several promising models from other disciplines that can be applied to economics – for instance, epidemiological models can be used to simulate banking crises. They choose to focus on agent-based models, however.

Agent-based models seek to represent the diversity of different agents, rather than trying to aggregate an entire population into a single actor. They draw on Monte Carlo simulation, a probabilistic method originally developed by physicist Enrico Fermi as a means of estimating the movement of particles.

In these models, agents interact with one another and sometimes their environment as well, following a set of behavioural rules. “The important feature of ABMs is that they explain the overall evolution of a system by simulating the behaviour of each individual agent within it and then explicitly combining their micro-level behaviours to give a macro-level picture,” Haldane and Turrell explain.

ABMs are one tool in a wider toolkit. The authors note they can be used for forecasting, but dynamic factor models or machine learning are likely to be superior. Instead, ABMs are more effective at conditional forecasting, assessing the likely impact of different policy approaches. DSGE models are used in the same way.

As with any model, there are trade-offs – ABMs sacrifice some internal consistency (allowing agents to be non-rational) for better external consistency. They are also less straightforward to explain than the simple three-equation model at the heart of DSGE models, as they rely on complex interactions. Furthermore, while it is theoretically possible to solve ABMs numerically, in practice it is extremely difficult to do so – the same problem Fermi was wrestling with. Hence the results are probabilistic.

In physics, such complex simulations are often a starting point for exploring hypotheses, the authors note: “Once a specific effect is identified within the complex simulation, a purely theoretical model (or a much simpler numerical model) is built to explain its salient features.”

The authors suggest several research questions remain for economists looking to maximise the potential of ABMs, including what are the most realistic behaviours to include, how should information flow between agents, and in what situations will ABMs and DSGE models yield similar or different results.

“Challenges aside, ABMs are a promising complement to the current crop of macroeconomic models, especially when making sense of the types of extreme macroeconomic movements the world has witnessed for the past decade,” Haldane and Turrell write.

  • LinkedIn  
  • Save this article
  • Print this page  

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact [email protected] or view our subscription options here: http://subscriptions.centralbanking.com/subscribe

You are currently unable to copy this content. Please contact [email protected] to find out more.

You need to sign in to use this feature. If you don’t have a Central Banking account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here: