Models need to be much more adaptable if they are to reflect the real world in a meaningful way, David Hendry and John Muellbauer say.
Both are professors of economics at the University of Oxford, and both specialise in econometrics. They are also highly sceptical of the use of rigid, structural models, such as the dynamic stochastic general equilibrium (DSGE) models widely used by central banks.
“Before the global financial crisis, macro[economics] was in a really terrible state, because the macro models central banks were using left out the most important things, namely credit, asset prices, housing – all those things that really mattered,” says Muellbauer.
Hendry and Muellbauer sat down with Central Banking for the second episode of the Rewiring Macro series.
They believe economic modelling needs to account for the constant shifts of the economy – which is not achievable with rigid models such as DSGE, which list equilibrium among many assumptions. In reality, economies are “highly non-stationary”, says Hendry – that is, they may constantly move towards equilibrium, but are knocked off course by one shock after another.
“We need models that are much more adaptable if we’re going to have a macroeconomy that matches reality to any great extent,” says Hendry.
Muellbauer allows that heterogeneous-agent DSGE models might be useful for thinking about theoretical questions, but he cautions against using them to set policy. “Policy models need to be much more embedded in time series data that takes full account of the rich structure of the economy… To get a tractable, useful policy model, the DSGE approach is just not going to help you,” he says.
Like David Vines, who discussed the need to reform DSGE models in the first episode, Hendry and Muellbauer see potential in agent-based modelling. Future episodes will explore these ideas in more depth.
01.00 – Have macroeconomists gone astray somewhere?
04.16 – Historical background
09.25 – Adapting DSGE models
14.50 – Empirical alternatives
25.50 – Agent-based modelling
30.50 – Teaching models to undergraduates