–Hugely Difficult For BOE To Put Banks, Defaults In Core Model
–Experts: Old BOE Model Struggled With Credit Shock, So Will New One
By David Robinson
LONDON (MNI) – Despite calls for a radical rethink of its macro
economic models, the Bank of England’s development of its new core model
will be an evolution rather than a revolution – with the focus on
simplification and integration with non-core models.
The BOE’s long drawn-out development of the new core macro model,
called Compass, has been as much, if not more, of an exercise in
developing the processing and software infrastructure for it to
facilitate interaction with other non core models than reworking the
model itself, Market News understands.
In the wake of the credit crunch there were calls for a fundamental
overhaul of macro models, to get them to incorporate the key elements of
the crunch – banks, credit channels and defaults. Experts say, however,
that incorporating such things into the BOE’s core model would risk
hugely complicating it without increasing its forecasting accuracy.
The BOE’s approach on Compass has, instead, been to work on a
simpler core model from the same school as the current model while
developing the satellite models to try and capture some of the new
thinking on the importance of such thing as credit channels and
financial accelerator effects.
Charles Goodhart, a leading monetary economist and former BOE
Monetary Policy Committee member, highlighted the limitations of the
BOE’s core model taken on its own.
“As a monetary economist it didn’t have anything I am really
interested in,” Goodhart told Market News.
The BOE’s core model “didn’t have any financial frictions in it”
and “there were no banks,” in so far as banks were simply represented as
channels of the central bank’s monetary policy, Goodhart said.
However he, and others familiar with the central bank’s model,
highlight the difficulties that would be encountered if they tried to
replace the current core BOE Quarterly Model, BEQM, with one that tries
to factor in the key components behind the credit crunch.
BEQM is, in econometrics jargon, a dynamic stochastic general
equilibrium model and Compass also looks set to be a DSGE model. Such
models are pretty standard. For example, the European Central Bank’s
New-Area-Wide Model (NAWM) is a large scale DSGE model.
“There is a big argument in macro about the extent to which DSGE
modelling has caused the profession to miss a trick. DSGE models worked
well when we lived in a time when financial markets worked,”
Warwick University’s Michael McMahon, who worked at the BOE and ECB, said.
McMahon is well aware of both the limitations of what current core
central bank models can do and the formidable technical challenges of
trying to build into them such things as the amplifying effects of
credit and financial market shocks, or “financial accelerators”.
Another former BOE economist told Market News a troubling aspect of
working with the model was asset prices, such as house prices, were
modelled outside the core and no matter what happens to them the linear
relationships in the core were assumed to hold good.
As McMahon said – in such a linear model – if house prices move,
say, plus or minus 5% the model just assumes the impact of a 50% move
will be ten times that of the 5% one.
But a 50% move “clearly might trigger other effects which didn’t
happen at +/- 5%,” he said.
“If the old (BEQM) model missed that, the new (Compass) model will
miss that as well,” McMahon added.
Factoring in defaults into the core model when asset prices fall
sharply or credit spreads surge wider might seem an obvious improvement,
but Goodhart says it would be highly problematic.
“The problem is defaults are a very non-linear development,” as
“defaults come in waves” and it is very difficult to predict when they
will occur, Goodhart said.
Factoring them in “inevitably makes the model greatly more complex
and it doesn’t make the model much better at forecasting,” he added.
The probability of a wave of defaults at a given time is very low
and “these things are tail risks and tail risks are essentially
unforecastable,” Goodhart said.
We already have a fair idea of what the new BOE model will look
like, in part because one of its guiding lights, Richard Harrison of the
monetary analysis division, has published the thinking behind it.
Harrison’s working paper set out a “linearised DSGE model” which he
said was as a simplified BEQM core model with a raft of frictions.
Harrison said “the model is very similar to the DSGE models of
Christiano … and Smets and Wouters,” putting it firmly in line with
the latest crop of central bank DSGE models, not least the ECB’s.
Two of the economists Harrison cites are Frank Smets, Director
General of the ECB’s research arm and Raf Wouters, who heads up the
econometrics arm of the National Bank of Belgium. Smets and Wouters’
joint, award-winning academic work is integral to the ECB’s model.
While the new core model may end up in the modern central banking
mainstream, the BOE will look to incorporate alternative economic
thinking into its suite of satellite models.
McMahon says the BOE is very much aware of the alternative
thinking, typified by the Kiyotaki-Moore model of credit cycles which
showed how small economic shocks can be amplified by credit
restrictions. These effects can be incorporated into satellite models
without hugely complicating the core one.
Asked about the new model last August, BOE Chief Economist Spencer
Dale said the aim was to make the core work better with the
satellites, saying “Our objective was to actually produce a smaller
model which is more tractable.”
He said the new model project should be seen not just “in terms of
building a new central model but rather building a platform on which you
can incorporate the insights from a whole range of different models.”
The new model is expected to be in place in the not too distant
future but the BOE is not giving out any a target date. with the
upcoming February Inflation Report forecast clearly too early for it.
Both BOE insiders and expert outsiders continue to express
exasperation with criticism of the BOE’s model which should, instead, be
directed at the assumptions that go into it.
BOE Governor Mervyn King’s mantra is “forecasts are not made by
models, forecasts are made by individuals,” that is MPC members.
The BOE came under fire for underestimating inflation, in
part because it underestimated pass-through from the fall in sterling.
Danny Gabay, a former BOE economist and now director at Fathom
Consulting, which works with the BOE model, said at one stage after the
sharp sterling decline of 2007/08 that the Inflation Report projections
appeared to be assuming no pass through into prices.
“That, I don’t think, is a modelling problem, it is an assumption
problem,” he said.
–London newsroom 0044 20 7862 7491; email: drobinson@marketnews.com
[TOPICS: M$B$$$,M$$BE$,MT$$$$]