Post-Autistic Economics Network
from post-autistic economics newsletter :
issue no. 6, May, 2001
Back to Reality
Tony Lawson (Cambridge University)
this essay appeared in Le Monde, 27 March2001]
In recent months a number of French students,
joined by some of their lecturers, have
initiated a debate on the state of modern economics. The debate turns on the
which research methods are appropriate for the investigation of economic
reality. As so
often in the past, a French debate has provoked an international response.
the importance of this debate to the future of economics it is essential to
be clear about
what is at issue, especially in these pages where the debate began.
Simply put, the message from the students is that there is insufficient
pluralism in the
modern economics faculty. In particular, there is a widespread insistence on
of just one set of methods: those of mathematical modelling.
A standard response to this observation, one which is also found in recent
pages of Le
Monde, and the one I would like to address
here, and is that this emphasis is unavoidable
just because economics needs to be scientific, where being scientific
use of mathematics.
When stated as starkly as this I think
it will be seen that the response is inadequate.
Most clearly it begs the question as to why economics needs to be scientific.
actually, its central deficiency is to presume unquestioningly that a science
uses mathematics. Such a presumption is false. What is more, a little
the nature of natural science suggests that there is every reason to suppose
an economics almost devoid of mathematics can yet be scientific in the sense
science. Thus the heading in Le Monde of 31/10/2000: “Les mathématiques,
nécessaire mais pas suffisante aux sciences économiques”
is actually quite wrong. Let
me briefly elaborate.
I take it we all agree with the French
students that illuminating social reality is the
primary objective. Certainly, I find few, if any, commentators rejecting this
The point here is that mathematical methods of the sort used by economists
(as with any methods) useful to the task of illuminating reality only under
conditions. Specifically, the usefulness of the sorts of mathematical
question is restricted to systems in which event regularities(deterministic
occur. Thus for those who suppose that science means using mathematics, the
that economics can and ought to be scientific is, in effect, a claim that
prevail in the social realm.
MauriceAllais, one of France’s great economists, has formulated this claim
when he writes:
essential condition of any science is the existence of regularities which can
analysed and forecast. This is the case in celestial mechanics. But it is
also true of
many economic phenomena. Indeed, their thorough analysis displays the
regularities which are just as striking as those found in the physical
sciences. This is
why economics is a science, and why this science rests on the same general
and methods of physics” (Allais, 1992, p.25).
is actually quite wrong in both aspects of his claim. Econometricians
repeatedly find that their supposed correlations are no sooner reported than
found to break down; social event regularities of the requisite sort are hard
to come by.
And, more to the point, it is just not the case that event regularities are
science. Let me defend this claim.
Actually, although the successes of
natural science are widespread, event regularities of
the requisite sort are rather rare even in the natural realm; out side
they are mostly restricted to situations of well-controlled experiment.
of the results of well-controlled experiments are successfully applied
controlled experiment where event regularities are not at all in evidence.
We can make sense of these observations
only by realising that the aim of the controlled
experiment, and of science more generally, is not the production of an
se, but the identification of an underlying mechanism that can account
for it. Gravitational
forces may give rise to an event regularity in an experimental vacuum, but
forces continue to act on autumn leaves wherever the latter may fly, and help
us to send
rockets to the moon.
It is an understanding of the
mechanism not the production of an event regularity that is
the essential goal here. The controlled experiment constitutes a human
aimed not at producing an event regularity for its own sake but at
(or testing a theory about) an underlying mechanism.
Medical researchers are not interested
in correlating the temperature of a patient with
the intensity or location of spots on the patient’s body, but with
counteracting) the virus or cause behind the symptoms.
In short, if there is a unifying feature of (pure)science, it is the search
for causes behind
phenomena regarded as of interest. If there is an essential component common
successful science it is this movement from phenomena at one level to their
in terms of causes lying at a deeper one. Mathematics is useful in the few
experimental) cases where surface phenomena are correlated. But science goes
its work of uncovering causes even where correlations in surface phenomena
are not to
So science is quite feasible in
economics. It entails identifying the causes of phenomena
of concern, say of high levels of unemployment or poverty. If mathematical
useful to this process, then so much the better. The central point, though,
is to recognise
that, whether or not they are useful, mathematical modelling methods are not
for any research process to qualify as being scientific in the sense of
My Cambridge colleague Professor Amartya Sen was correct when recently in Le Monde
(31/10/2000) he observed that mathematics is not a unique foundation of
science. In fact it is not a foundation
of economics-as-science at all.
Actually, it is my own view that we
can go further than this. We have good reason to
suppose that the scope of relevance of mathematics is very limited indeed in
realm. For example, it can be demonstrated that not only the poor success
modern economics, but also the phenomenon of modern economists repeatedly
assumptions known to be wildly false, are due to mathematical methods being
employed where they do not fit. These are amongst the assessments I defend at
in Economics and Reality (Lawson,1997).
But they not essential to the points being
made by the French students, and I put them aside here. The students’
only that, in modern academic economics departments, mathematical modelling
pursued for its own sake. They argue, and I agree, that we should start with
(or at least
not neglect insights concerning) the nature of reality. The point is not to
mathematical methods a priori, but
to use such methods as and when appropriate.
One final point. I have set out a
conception of science that some will contest. It is
possible indeed that it will prove inadequate. Or time may show that my
about the relevance of mathematical modelling for economics is unfounded. All
knowledge is fallible, after all. But to recognise that any argument or claim
turn out to be wrong is to acknowledge, at the same time, a need for a
non-dogmatic, indeed more pluralistic, approach in the academy.
This, of course, is just the first and
most fundamental point of those of us who are
unhappy with the state of modern economics. The objective is not to replace
by another. Certainly it is not an a priorir
ejection of the use of mathematics in
economics. Even less is it a rejection of the possibility of economics as
And nor is anyone suggesting an abandonment of standards of rigour in the
relevance. Rather, the goal is simply to open up the economics academy to a
intellectual orientation, allowing, in particular, the combining of high
research with a return to variety and greater(albeit critically informed)
pluralism in method.
Allais, M. (1992) "The Economic Science of Today
and Global Disequilibrium", in Baldassarri M.
Global Disequilibrium in the World
Economy, Basingstoke: Macmillan.
Lawson, T.,(1997) Economics and
Reality, London: Routledge.
Tony Lawson, (2001) “Back to Reality”, post-autistic
economics newsletter : issue no. 6, May, article 2. http://www.btinternet.com/~pae_news/review/issue6.htm