The
reputation of macroeconomics took a battering in the wake of the global
financial crisis after failing to predict the
great recession. Although much of the criticism by outsiders is misplaced,
there are some grains of truth and many academic economists would agree that
there are many areas where economics needs to improve.
This collection of papers from
the Oxford Review of Economic Policy looks at the state of macroeconomics today
and provides a range of opinions from leading macroeconomists. More
importantly, it shines the spotlight on those areas where economics can be seen
to have failed and offers some suggestions about how to take us forward (the
papers are not particularly technical and as such are relatively accessible.
Credit should also go to the publishers, Oxford University Press, for taking
this volume from out behind the paywall).
David Vines and Samuel Wills make
the point that macroeconomics has been here before – in the early 1930s and
again in the 1970s, and both times the discipline evolved to try and make sense
of changed circumstances. But in order to identify what has to change, we need
to know where we are and what is wrong. At the centre of the debate stand New
Keynesian Dynamic Stochastic General Equilibrium (DSGE) models, which form the
workhorse model for policy analysis.
The general consensus is that they are not
fit for purpose – a point I have made before (here and
here).
Such models are based on microfounded representative-agents – a theoretical
approach which postulates that there is a typical household or firm whose
behaviour is representative of the economy as a whole. I have always rather
struggled with this approach because it assumes that all agents respond in the
same way – something we know is not true in the case of households given
differing time preferences, depending on age and educational attainment. An
additional assumption that underpins such models is that expectations are
formed rationally – something we know is not always true.
Thus
the consensus appears to be that these two assumptions need to be relaxed if
macroeconomics is going to be more relevant for future policy work. You might
say that it is about time. Indeed it is a sad indictment that it took the
failure of DSGE models during the financial crisis to convince proponents that
their models were flawed when it was so obvious to many people all along.
In
order to understand this failure, Simon Wren-Lewis offers an explanation as to why this form of thinking became so predominant to
the exclusion of other types of model. He argues that the adoption of rational
expectations was “a natural extension of the idea of rationality that was
ubiquitous in microeconomic theory” and that “a new generation of
economists found the idea of microfounding macroeconomics very attractive. As
macroeconomists, they would prefer not to be seen by their microeconomic
colleagues as pursuing a discipline with seemingly little connection to their
own … Whatever the precise reasons, microfounded macroeconomic models
became the norm in the better academic journals.” Indeed, Wren-Lewis has
long argued that since academics could only get their work published in top
journals if they went down this route, this promoted an “academic capture”
process which led to the propagation of a flawed methodology.
Wren-Lewis
also makes the point that much of so-called cutting edge analysis is no longer
constrained to be as consistent with the data as was once the case. He notes
that in the 1970s, when he began working on macro models “consistency with
the data was the key criteria for equations to be admissible as part of a
model. If the model didn’t match past data, we had no business using it to give
policy advice.” There is, of course, a well-recognised trade-off between
data coherency and theoretical consistency, and I have always believed that the
trick is to find the optimal point between the two in the form of a structural
economic model. It does not mean that the models I use are particularly great –
they certainly would not make it into the academic journals – but they do allow
me to provide a simplified theoretical justification for the structure of the
model, in the knowledge that it is reasonably consistent with the data.
Ultimately
one of the questions macroeconomists have to answer more clearly – particularly
to outsiders – is what are we trying to achieve? Although much of the external
criticism zooms in on the failure of economists to forecast the future, what we
are really trying to do is better understand how the economy currently works
and how it might be expected to respond to shocks (such as the financial
crisis). Olivier Blanchard
believes that “we need different types of macroeconomic models for different
purposes” which allows a continued role for structural models, particularly
for forecasting purposes. Whilst I agree with this, I have still not shaken off
the conviction, best expressed by Ray Fair back in 1994 (here
p28), that the structural model approach “is the best way of trying to
learn how the macroeconomy works.” Structural models are far from perfect,
but in my experience they are the least worst option at our disposal.
No comments:
Post a Comment