ECONOMETRIC MODELLING: THE “CONSUMPTION FUNCTION” IN RETROSPECT*
| DOI | http://doi.org/10.1111/j.1467-9485.1983.tb01014.x |
| Published date | 01 August 1983 |
| Author | David F. Hendry |
| Date | 01 August 1983 |
Scoffsh
Jouroloffolificd Economy,
Vol.
30,
No.
3.
November
1983
Q
1983
Scottish
Economic
Society
ECONOMETRIC MODELLING:
THE
“CONSUMPTION FUNCTION” IN
RETROSPECT*
DAVID
F.
HENDRY?
Nufield College, Oxford
THE
NON-ECONOMETRICIAN’S LAMENT
As
soon
as
I
could safely toddle
My parents handed me a
Model.
My brisk and energetic pater
Provided the accelerator.
My mother, with her kindly gumption,
The
function
guiding my
consumption
;
And every week I had from her
A lovely new
parameter,
With lots of little
leads
and
lags
In pretty parabolic bags.
With optimistic
expectations
I
started on my
explorations,
And swore to move without a swerve
Along my sinusoidal curve.
Alas!
I
knew how
it
would end
:
I’ve mixed the cycle and the trend,
And fear that, growing daily skinnier,
I
have at length become
non-linear.
I
wander glumly round the house
As
though
I
were
exogenous,
And hardly capable
of
feeling
The difference ’tween floor and ceiling.
‘Eighteenth Annual Lecture of the Scottish Economic Society delivered at the University
of
Stirling on
5
February
1983.
The Society acknowledges with gratitude the financial support
provided by Shell International for the Annual Lecture.
?This
research was supported in part by Grants from the Social Science Research Council to
the MIME Programme at the London School
of
Economics and by HR8789
to
Nuffield College.
I
am
indebted
to
Frank Srba and Neil
Erinson
for the calculations reported below.
The
paper was
prepared at the invitation
of
the Scottish Economic Society, the concluding section then being
modified
to
clarify issues arising in discussion. Helpful discussions with Neil Ericsson, Robert
Marshall and John Muellbauer are gratefully acknowledged.
Date
of
receipt of final manuscript:
20
May
1983.
193
194
D.
F.
HENDRY
I
scarcely now, a pallid ghost,
Can tell
ex-ante
from
ex-post
:
My thoughts are sadly
inelastic,
My acts incurably
stochastic.
D.
H.
Robertson
3 September 1952
(italics added)
(from:
The Business Cycle
in
the Post-War World,
ed.
E.
Lundberg,
Macmillan, 1955)
I
INTRODUCTION
The main aim of this paper is to exposit a number
of
concepts relevant to
econometric modelling
of
economic time-series data by reappraising the
model developed in Davidson
et
ol.
(1978) (denoted DHSY below). On the one
hand, this should help to clarify their investigation, which was conducted on a
rather intuitive basis as a “detective story”, without formal analyses of several
important notions and model types. Such analyses have appeared only
recently. On the other hand, the relatively widespread use of the DHSY model
(see the comprehensive study by Davis, 1982) suggests using it to illustrate the
ensuing developments in econometric “method” and
so
generate a greater
understanding
of
both the terminology and the conceptual basis
of
many
recent modelling ideas.
As
the preceding poem reveals, however, the language
itself is not recent;
plus ga change.
. .
.
The technical basis for the present
discussion is presented in Hendry and Richard (1982), (1983) who also provide
more complete bibliographic perspective.
A
distinction of importance in what follows is between
eoaluating
and
obtaining
empirical econometric relationships. These two aspects are certainly
linked in
so
far as evaluation techniques help to exclude some potential models
and may even comprise the main basis
of
various methods aimed at locating
useful data descriptions. Nevertheless, a satisfactory evaluation outcome
cannot be sufficient by
itselffor
establishing a useful empirical equation, and at
best is only necessary.
Precisely how one should
or
even could obtained useful empirical models is
not obvious.
As
Haavelmo (1944, p. 10) expressed the matter, “It is almost
impossible, it seems, to describe exactly how a scientist goes about construct-
ing a model. It is a creative process, an art”.
.
.
.
Thus, important elements
include inspiration in the formulation
of
both relevant theory-models and
their data analogues, as well as serendipity. Whereas formal evaluation is
mechanistic, obtaining is almost the antithesis.
To
quote a well-known
illustration, rejecting a null hypothesis against an alternative does not entail
the validity of the latter.
As
stressed by Cross (1983)-citing Duhem-
To continue reading
Request your trial