===== PAGE 1 =====

OSKAR MORGENSTERN 365

Does GNP Measure Growth And
Welfare?

PRIMITIVE IN THE EXTREME AND CERTAINLY USELESS
By OSKAR MORGENSTERN, Economist, Professor of Economics at New York University
Delivered at New York University, New York, New York, February 13, 1975

HE question posed by the title to this lecture raises check, etc. But to observe “growth” or “utility” or “welfare for a

a number of deep problems of economic science. The whole nation is an entirely different matter. Surely people are in-

question seems simple, yet in these few words — volved, and cars that can be counted, or monies changing hands and
measurement, growth, and welfare — culminate the problems of so on, but to state how the economy grows and whether welfare
large areas with which economists have been concerned for cen- changes and by how much are entirely different matters. They ought
turies. How nice it would be then if one could report that now we to be approached with caution, respect, indeed with trepidation.

have one single concept and, indeed, one single number which could When we mention such words we know that they mean very
summarize variations of growth and welfare. As we shall see this un- different things to different people. In fact, for many they may mean
fortunately is not the case. nothing concrete or tangible. They are abstractions, possibly of a

To begin with the problem of measurement: All sciences have to higher order. We feel that there is some reality touched, but we also
come to grips with that problem. It is an extremely difficult one and know that it is a long way from that state of affairs to truly scientific
each science has its own troubles. When a measurement in any field objective knowledge. What we need in order to observe, and
has become possible this is rightly hailed as a great achievement, asa hopefully to measure, are concepts. There is no escape from that. No
step forward towards new discoveries and innumerable applications. matter how primitive our approach may be, a concept is involved.
Even to have a precise measure of time is difficult: clocks are a late Good, useful concepts are hard to come by.
development in human history and to this day we try to make them Let us first look at “growth.” That is clearly a notion that applies
more and more accurate, because so much depends on them. So itis to an organism. The human body grows, stops growing, and even-
not surprising that when it was recently reported that the rotation of tually dies. The body has many parts and many functions. It grows
the earth — our finest measure of time — may have slowed down not only as a whole, but parts of it grow at different rates and not for
during the last year by one second, this caused great interest in ever as already analyzed by Galileo, and the whole is of un.
physics and astronomy. From this reaction we also get an idea of the imaginable complexity which modern biology reveals to us in a rapid
high standards in those sciences. In other fields we measure blood though still partial manner. In the whole process of organic growth
pressures, barometric pressure, the composition of blood, there is control, genetic planning and unified function. Can the
temperature, the speed of light, etc., etc., sometimes crudely, economy be compared with this so that we could transfer the notion
sometimes with extraordinary precision. of organic growth from this area to our field? One thing is certain:

Although in the physical and even in the biological sciences a the economy is also of the highest complexity. Consequently one
great and firm tradition has built up over the centuries the matter of must expect that even an adequate qualitative description is cor-
observation and measurement is never closed. Science is never respondingly difficult.
finished; rather, for all time, science is only an approximation to the I find however that very few men, even few economists, or should
underlying reality. It may surprise you that in spite of this great I say regretfully, especially economists, have a real appreciation and
tradition Einstein, as he has more than once remarked to me, has understanding of the immense complexity of an economic system.
said: ‘Most scientists naively think they know what they should Now I have used the word “system” but it is not even clear whether
observe and how they should measure it.” And he had the natural the word “system” is appropriate because we do not understand fully
sciences in mind! the organizing principles that make the economic life of a nation

How then is the situation in the social sciences, in particular in possible. One need only observe what happens in an ordinary per-
economics? In some manner we seem to be more fortunate than the son’s day and what is considered to be absolutely obvious and nor-
natural sciences. Nature shows itself to our senses essentially in a mal. We take a bus in the morning, we take a train, we go to stores,
qualitative way and to get to numbers of high precision requires we buy, the money is accepted with which we pay, or we pay by a
great effort and is a formidable achievement. But in economics we check which will be presented to an anonymous institution, possibly
can count right away, for example, the number of inhabitants of 2 far away, transmitted by mails about which we have no control
town or a country, the number of motor cars, of checks cashed, of whatsoever; the stores order and have things in stock; they set prices
tons of steel produced, of ... well of almost anything — or so it on the basis of the expectation of what competitors might be doing;
seems. Since it looks to be so simple to observe and count, there is in we are buying on the same basis of expectation as to what our needs
economics — as well as in other social sciences — no tradition might be in the future and how they will be met; we spend our
which would enforce high standards. Counting seems natural and money today because we expect that our salaries or incomes from
what is counted is readily accepted. But that is true only up to the other sources actually will be paid at certain dates in the near or
point where broader notions are encountered, such as “growth” and more distant future, etc., etc. In all this, there is no central genetic
“welfare.” Or even more basic, the question of what is “value” and regulation, nobody who plans for all; there is no one to whom
“utility” promptly causes great difficulty. Suddenly we see that we everybody is responsible. We carry a great deal of information in
have to have sharp concepts, that we must ask questions to which we our heads, and that which we do not carry we can find in newspapers
normally get only qualitative, not right away quantitative, answers. which gather information and are printed, again motivated only by
There is little doubt what an individual is, or a motor car, or a their own self-interest, namely, to write, to publish, and to sell with


===== PAGE 2 =====

366 VITAL SPEECHES OF THE DAY

profit. At present, at this very moment, raw materials are being Now I want to point out only one of its quantitative features: It is
produced, for which there is no conceivable way of telling how they expressed by a single number, a so-called scaler. We discussed
will be finally molded, into which kind of finished products they will organic growth of, say, a human body. Would anybody in his senses
be turned. A steel producer has no idea whether his steel will be used imagine that there could be a single scalar number which would
for the making of tanks, or ships, or paper clips. adequately describe the development of a human from babyhood to
And yet all this works and works miraculously well, although of maturity to old age: the growth of the body. of the mind. ot
course it is easily subject to great disturbances. The astonishing fact capabilities? The idea is so grotesque and ludicrous that we can dis-
is not that the thing does not work well, but that it works at all. It is miss it from the outset. We have just seen that the economy too is of
only when we realize the complexity of the economy that we begin high complexity and that therefore its description. or rather its
to see, and possibly understand, how dangerous it is to interfere in changes, could be given and measured -— accurately. without the
these matters. The economic system is in addition subject to great slightest error of measurement — by one scalar number is equally
changes, caused partly by technology, by strictly political events, absurd. If one could design a method whereby changes are at least
partly by changes in the desires and wishes of the final consumers. recorded by a vector — i.e., a number that has many components -
There is thus more than one uncertainty element which governs it could be a very different matter. I know of no such effort. Ot
course a ‘vector’ is a more sophisticated concept and would not be
The human body, or any living organism, is in many ways a very useful for the simple and trivial interpretations which the single,
much simpler matter. However, if one knows anything at all about scalar, unit makes so easy. Just to show the absurdity and limitation
the body's physiology, one would find that statement horrendous, of this popular alleged "concept consider the following :
because one can hardly imagine anything more finely tuned and Anything that leads to a transaction in monetary form, which
more complicated than let us say, the human body, not to mention is where goods and services change hands against money, is recorded
the brain, which without any doubt, is the most complicated thing in as positive. no matter what is being sold, it enters GNP. It may have
the whole universe of whose existence we are aware, and whose been sales of goods already stocked. it may have been a car just com-
function we clearly do not understand, although we have a tremen- ing out of a factory: it does not matter. Neither does it matter what
dous amount of knowledge of it already. The fact is that the human it is: Atomic bombs, drugs, cars, food, aesthetic pollution by new
body has one unique purpose, namely to stay functioning, and to stay billboards, . . . you name it. Clearly that goes against common sense.
alive. We cannot say that the economic system has any such clear Why should all products and services be treated alike: If 1 do not
purpose. like more nuclear weapons why should I accept a measurement that
In the light of such observations, one should become extremely includes them as part of the "growth" of the economy’ Of course
modest in making proposals for policy. In general, I would say that one could argue that one is only interested in transactions. But then
unless we are reasonably sure that we know what the consequences one would have a great deal of explaining to do how more transac-

everything.

of new policy measures will be — for example of new taxes in- tions can possibly be related to “welfare.” Does the uncontrolled in-
troduced, prices regulated, etc., etc. — we should leave things alone. crease of cancer cells in a child mean growth’:
One interferes only if one believes that one understands what the There are other equally well-known difficulties. Many services

consequences of the interference are. The same is true of medicine: are rendered and many goods are produced that never enter a
It took a long time to come from the witch doctor to the modern market. Thus they escape GNP. As has been noted by many. if
brain surgeon. While presumably the human physiology. or that of housewives were being paid by their husbands GNP would rise
any animal or any other biological entity stays practically the same although there would not be one iota of difference in production or
throughout centuries or millenia — although subject to evolution services. There are many other similar situations.
over hundreds of thousands of years — the economic system, to Another trouble with the GNP concept is that it measures, or
make things still more difficult, is constantly being changed. es- rather expresses, as positive also the malfunctions of the economic
pecially by technology. Technology is an interference from the out- system or society. To wit: if we are stuck in one of the thousands of
side; it is being absorbed in a manner which those who bring it into traffic jams, if airplanes are stacked and cannot land on schedule. it
economic existence think to be profitable for them to do but whose fires break out and other disasters occur, that require repair — up
global effects they neither understand nor care about. Clearly, this goes he GNP. More gasoline is used, fares go up, overtime has to be
compounds the difficulties of policy, because what might have been paid, and so on. It would be difficult in any other science to find a
a suitable measure to achieve desired results at one time may no ‘measure’ which tells simultaneously opposite stores of the func-
longer have any validity under present or future circumstances. It tioning of a complex system in one single scaler number! If we
takes a long time to develop a scientifically acceptable new idea of merely improve the scheduling of airplanes and stagger the times ot
policy and while it is being born, if that happens at all, new features automobile traffic, and nothing else is changed -- down goes GNP!
appear in the economic life. For example, Keynesian policy ideas of It goes up, on the other hand, if industry pollutes the air and we
the 1930's are now being adopted by the U.S. government at a time create other industries which remove the polluting substances.
when they have virtually no applicability whatsoever. So we see that there is real trouble with the basic underlying no-
Thus the “growth” of an economy is a very different matter from tion of GNP. It is not an acceptable scientific concept for the pur-
the growth of an organism. The economy has neither a beginning poses it is used. The fact that it violates common sense might be con-
nor an end. In addition, as already said, the economy certainly sidered not too critical. There are, after all, many concepts in physics
changes. So it is perhaps not surprising that apart from some highly which common sense could never create or judge, or might reject,
technical and abstract models of economic expansion — which is not such as, for example, “curved space.” But these concepts are the
the same thing as growth — there simply does not exist a generally product of powerful theories and are needed when a scientific field is
accepted scientific concept which would give us the basis for a already well developed. But the GNP action is nothing of this kind.
reliable numerical measurement of the raze of growth. It expresses a trivial idea which is clearly accessible to scrutiny by
But there is GNP, gross national product, i.e., the turnover in an common sense. Therefore its current, indiscriminate uses are suspect
economy in a unit period, a hallowed notion in contemporary and it is certainly questionable that it should be used to tell us about
economics. It is used with abandon. I shall say more about it later. growth and welfare.


===== PAGE 3 =====

OSKAR MORGENSTERN 367

At this point it is proper to recall Einstein's remark about the fact ment that the economic value of an object is only what you can sell it
that it seems to many obvious what one should observe. Surely trans- for: “Res tantum valet quantum vendi potest,” which is certainly
actions occur all the time in the economy, but that does not mean true for shares on the stock market, though there is also the assertion
that they offer the proper way to describe the functioning of the of an “intrinsic” value which is supposed often to differ in both
economy. Perhaps only some transactions may give us the desired in- directions from the sales value. Though all this may seem to be a
formation, only a selection made on the basis of a powerful theory. confusing situation there is today no doubt that — apart from the
But what we see is that the corrections — or better changes — of few remaining adherents of the superseded labor value theory —
GNP figures that are made by elimination of the effect of seasonal utility attributed by individuals to goods and services is all that
variations, of price changes, etc., therefore obtaining a stable basis, matters. Utility is based on individual preferences and these are
eliminating the influence of distortion caused by inflation, in no way related to the objective, technical characteristics of the desired goods
touch the fundamental issues and objections. and services. The individual can compare the utilities he expects to

GNP is a global notion. It is undifferentiated. It falls into the derive from them even though the goods are all different. One can
pattern of modern macro-economics where the attempt is made, even find out that individuals can construct numbers for their
perhaps largely under the influence of Keynes, to relate, say, the preferences, so that they are better guided in their attempt to obtain
total quantity of money in circulation to fota/ employment, total out- a maximum of expected utility by suitable choice and corresponding
put of industry, etc. It is tempting to do so and would simplify expenditure overtime of their money or other income. This utility,
economic reasoning enormously if one could discover strict in- derived from all goods and services, then is individual, personal,
terdependencies. Yet there is great danger in these efforts; for exam- welfare.
ple, the same increase in the quantity of money will have very But here we come to an end: Utility is strictly an individual
different consequences when it goes to consumers rather than to matter. Exactly the same bundle of goods will have a very different
producers. This is obliterated when one restricts oneself to the macro utility for different persons. So much is obvious in a gross sense and
entities. need here not be elaborated any further. But what does #or follow is

It is interesting to note that modern science goes in exactly the op- that we can compare utilities of different individuals. The fact is
posite direction: more and more, finer and finer distinctions are be- precisely that different individuals are different. Only if they were
ing made. First one has a molecule, than an atom, then the electron, identical, like the famous twins, would they have the same utility
then more and more elementary particles, even subparticles and only functions. Perhaps even with them there may be differences because
by these steps does one arrive at a better understanding of matter. Or who can look into their brains to make sure that there are no
in biology: there was only cancer: now we know that there may be differences in their wishes, preferences, ideas and ideals? Of course it
100 different kinds, each with different effects and possibly calling is likely that persons in a given income class have similar needs and
for different treatments. So it is everywhere. But GNP, as an alleged are concerned about the prices of similar bunches of goods. But if I
global measure, runs precisely counter to the spirit of modern science take a good from A, for whom it has some utility, and give it to B,
even on these grounds. for whom also it has some utility, it does not follow that I have

I'am tempted here to quote St. Augustine: “For so it is, oh my diminished A as much as I have improved B's position. Once more:
Lord God, I measure it, but what it is that I measure I do not utilities of different individuals are incomparable. There is now no
know.” known way to make them comparable, to find out objectively

When we now talk about “welfare” we are entering upon a field whether I transfer the same utility (for example by taxation) from
where other great difficulties arise. Economists have struggled with one to the other.
the problem of welfare for centuries. Countless many volumes have Economists have thought of one way which seems to give us in-
been written about it but one has not yet come to a resolution. It is formation about general welfare, which is what we are really in-
therefore likely that using a primitive notion such as GNP to terested in and what GNP is somehow expected to measure. It is the
measure that much disputed thing will come to nought. But first let notion of the so-called Pareto optimum which says the following: if
us examine briefly where the difficulties lie with respect to economic we look at a community and add a good to one party without at the
“welfare.” Most of us will think at first of our own, personal same time diminishing any other party of that community, then we
situation: possessions, income, stability of income, needs (as can say that the welfare of the whole group has increased. This
perceived), health, obligations to others, prices of the goods and ser- seems plausible and harmless and we find the Pareto optimum ex-
vices to be bought over some more or less specified period of time: in tolled and used in almost every textbook in economics. Yet how can
short our personal welfare is composed of many variables. Over time we find out whether this principle holds? Only by questioning —
some variables may go up while others go down. Do these since there is no objective way of making this statement. When does
movements cancel out? In some cases it is obvious; income may rise a person feel diminished? This might even happen when something
in the same proportion as the prices of the goods I want and so all is given to him that he does not want or cannot use (while for others
stays the same. But when some prices of relevance to us go up and it might be desirable and useful). Does A feel diminished when
others go down then it is not so obvious whether and how our in- someone else gets an addition? This can certainly happen even when
dividual welfare has been affected. nothing is actually zzken from him. It is an entirely different matter

Now men are different in age, incomes, wants, positions, desires. whether in a group one individual is given $1 or $1,000,000,
The difficulties arising here could be overcome by broad though formally the idea of the Pareto Optimum is preserved. Com-
classifications of people, by the law of large numbers and other such mon sense tells us otherwise. And how do we even find out whether
devices. But that still does not tell us about “welfare” because the someone benefits or is diminished? Certainly by interpersonal com-
principal attribute of the latter is in the last analysis what value or parison of utilities, which we know we cannot make. So we have to
utility people attach to their possessions and income. What deter- ask them. Do they tell us the truth? Necessarily? Always? May they
mines value has puzzled economists for ages. Is there an objective not be playing a game in order to extract some greater benefit? So
value applicable for everyone, in any circumstance, at any time? we see that this seemingly innocent and seemingly workable concept
How is value produced? What affects value? The answers have is applicable at best under severely restrictive conditions which may
ranged from the assertion that there are absolute values to the state- never be given in reality.


===== PAGE 4 =====

368 VITAL SPEECHES OF THE DAY

In reality we determine, of course, social preferences and act ac- today perhaps only 40 percent of activity involves physical output.
cordingly. We tax people and transfer income to others. We build the rest is “services.” And no one has come forth with good ideas as
public works, establish museums, run a military force and soon. But how to measure “productivity” of lawyers, doctors, teachers.
all this is the product of political decision processes, based on voting policemen, hospitals, scientists, musicians, actors, etc., etc. Yet they
or on dictates, with only vague ideas of what might be good for the all have some kind of “output”, conceivably even a “net” output . .
society and increase its “welfare.” But there is no strictly scientific But what is it and how do we compare these heterogeneous services

unchallengeable basis. There is power or at best persuasion. with each other? And though it would be good to differentiate what
The upshot of all this is that “welfare” is an elusive concept, that is hoped for is again a primitive, single, scalar number!
has, however, great intuitive appeal but means different things to Now let me turn to the way in which one deals with the numbers

different people and groups of people. It is a concept that slips purporting to measure GNP. Let us forget for the time being all that
through our net the moment we want to make it objective in a scien- was said in criticism of GNP. Let us merely look at the measurement
tific sense. I say this with all due respect to the many economists past itself.
and present who have given so much thought to this matter. Many Measurement is demanding, and accurate measurement ex-
valuable attempts have been made, but as so often in science, much  ceedingly so. It may interest you that even today we do nor know the
has to be discarded that was once considered to be acceptable. There Moon's precise distance from the Earth — though man has visited
is a great challenge here to economic science to find some day a the Moon! What is more, incidentally —- and I shall come back to
satisfactory expression and to tell us then also how to measure that in an application to economics — we cannot prove, in general.
welfare. the stability in the large of the Moon's orbit around the Earth. We
A promising development is the study of the so-called “social in- can only prove it "in the small.” For the whole proof we would need
dicators.” This movement recognizes that a positively valued social to know the Moon's behavior from the beginning to the end. Ex-
development depends on the simultaneous changes in many actly the same is true regarding the entire planetary system. Should
variables: An increase in production is not good if it is accompanied not these two considerations make us economists exceedingly modest
by more pollution, or a rise in income without at least no deteriora- and cautious when it comes to the discussion of measurement of
tion of its present distribution. Incidentally, note that when I speak complex economic-social situations and of the stability of the
of "deterioration of income distribution’’ I appear to be able to say economic universe?
in a scientifically objective way when a particular income distribu- But what happens in reality?
tion is “better” than another. This would again involve interper- Look at the mountains of economic statistics pouring forth con-
sonal comparisons of utility which we know have to be ruled out. tinuously from the Government and from business: Millions of
Yet common sense tells us that, to take an extreme case, if 90 per- numbers, immense detail as well as comprehensive aggregates and
cent of the national income goes to 5 percent of the population that sophisticated index numbers. Many numbers have been. so to speak.
country is hardly better off than when possibly even a smaller in- laundered, for example seasonal variations are eliminated in order to
come is distributed more equally among the inhabitants. This clash show the alleged true movement of some activities. Some numbers
of insight with our inability to make scientifically acceptable are obtained by carefully studied sampling processes; powerful

statements is most disturbing. statistical theory is used in many instances and sampling errors are
There are countless many interrelated factors on which welfare. carefully spelled out.

whether personal or communal, depends. Even to describe them. to But there is one characteristic that pervades all: we virtually never

enumerate them and to put them into a coherent picture is a difficult encounter the words “perhaps,” “approximately.” “about.”

task. Any differentiation which leads away from simple global ex- “maybe,” and so on, or see a sign saying: ""£ x percent. In that

pression is in the right scientific spirit which forces us to make more there is a great difference from the natural sciences where it 1s
and more distinctions. There is no doubt that such qualitative, customary, nay standard, to ask immediately what the error of
descriptive work has to precede any measurement, and it may give us observation might be. A notable exception is S.N. Kuznets who 1n
even now already the basis for rough comparisons of different states his valuable work on national income has shown that those figures
of the same country and some glimpses of comparing the welfare in may have an error in certain classes of income of even more than 20
different societies. percent. Though his studies have been gratefully accepted by the
This is what GNP is supposed to disclose to us — nay, more: economic profession there is little evidence that these errors are duly
even to measure with extraordinary precision! A poorly defined and considered and their implication is carried over to other fields. Ob-
hard to capture, yet exceedingly important phenomenon such as viously if there are errors in aggregates there are errors in their con.
welfare allegedly measured by another one, GNP, that records ponent parts.
nothing better than the total — 1 am almost inclined to say Error there must be. There is no perfect measurement, no matter
"hodgepodge'* — of transactions in the economy. The idea has no how good the underlying concept, no matter how fine the measure-
chance of finding any scientifically valid justification whatsoever. It ment tool. Some things seem easy to measure. ¢.g.. to count the
is once more, as in the case of growth that one demands of a single. number of inhabitants in a country, but it may surprise you that at
simple scalar number to produce wonders of measurement. the U.S. Census of 1950. about 5 million people — equivalent to a
Of course, attempts have been made to improve upon the GNP good size American city, say Chicago — were not counted as was
concept, for example by considering GDP (gross domestic product) later found out in a careful study.
in order to exclude foreign transactions, or attempts to arrive at some Other censuses may have been better or worse. In some Asiatic
measure of net output so that the "productivity of an economy’ can countries the population count is said to have errors of + 20 percent
be determined. Such efforts go clearly in the right direction, namely and who knows the population of China or of some African coun-
that one must make finer and finer distinctions however conceptual tries? For very gross comparisons this may not matter too much but
difficulties arise than too. For example, the notion of “productivity” the frequently and freely used international per capita income data
applies more easily to physical processes where we observe physical are usually meaningless unless countries are very similar to each
inputs and outputs. These can also be expressed in monetary terms other and the true intervals are short.
though this is not a simple matter. However, in the U.S. economy Errors vary enormously. In physics the finest measurement. 1 am


===== PAGE 5 =====

OSKAR MORGENSTERN 369

told, is that of the Ritchie constant whose error is an unbelievable very kindly distribute themselves in such manner that they all cancel
10 -* Yet cross-section data for nuclear reactors may have an error out?

of + 50 percent. But the reactors work! So this shows that the ex- The facts are, of course, quite the contrary: there must be errors:
tent of admissible error depends on the use for which the measure- after all, we live in a “stochastic universe,” which means that the
ment is needed. Of course, the smaller the error, the better. Reduc- world is to a large measure indeterminate. And that errors in
tion of error is usually expensive and how much one wants to spend economic observations would cancel out precisely — of that no one
on error reduction depends again on the importance of the use of the has given proof and no one will ever be able to do so in general.

measurement or observation. If a good number has to be combined I shall give two illustrations of calculations. Suppose you have two
with a poor one there is not much point in making the good number consecutive GNP numbers — whatever their absolute values be —
even better. such that the second one is 3 percent larger than the first one, assum-

Since errors range from very small to very large, is there a ing each one to be perfect. Now assume that the first is, however,
generally accepted standard? The answer is that one should always understated (i.e. has an error) by -1 percent, the second overstated
state honestly what error is involved, in whatever field. This is com- by +1 percent, then the rate of growth, instead of being 3 percent is
mon in physics and I regret to say almost totally lacking in in fact +5.08 percent. Quite a difference! But if, the other way
economics and the other social sciences. With this I refer to the basic around, the first is faulty by -1 percent and the second by 1 percent,
errors in variables, i.e., errors in the basic observations not to sampl- then the rate of growth is only 0.96 percent. What is reality? A one
ing errors and the like. There will have to be a change in attitudes, percent error one way or the other is tiny, by almost all standards.
in demands, in standards. No compromise is possible. It is difficult Recall that in physics, a paradigm for measurement, observations
to determine the various errors because, when the data were put with only a 3 percent error is frequently a very good and useful
down, no effort was made to discover the inevitable error. And vir- measurement. What happens to our growth rates if we repeat what
tually all these numbers are not the result of scientific approaches to was just done for 3 percent? In the first case, instead of the error-
the different phenomena. They originate from business transactions, free 3 percent basic change, we obtain +9.37 percent; in the second
from administrative regulations, etc. Prices are found in shops and case -3 percent. So the economy, as represented by an alleged error-
factories, export and import quantities and values are recorded on free change in GNP statistics, instead of having grown by 3 percent,
papers accompanying the commodities, investment figures and may have grown by 9.37 percent or decreased by -3 percent. This is
profits and total sales are found on balance sheets, incomes on in- the result of trivial calculations. Precisely for being so trivial they

come tax returns, and so on. show the grossness of the (tacit) assertion that the “true” rate of
To give but a few indications of what one finds: import and ex- change, or “growth” is allegedly +3 percent.
port figures of many countries, taken from the respective two Of course, if the errors in the two underlying GNP figures would

sources, often differ by 20, 30 or even 100 percent both for value or not only go in the same direction — i.e. being each either positive or
quantity. This applies even to the trade in gold! The total quantity of each negative — and in addition were to be identical, then nothing
money in circulation in the U.S. is supposedly not known to within would change, the rate would stay the same. But even if they should
perhaps $20 billion. Yet pick up almost any basic economic or pop- go in the same direction but were to differ from each other in con-
ulation statistic and you will find that figures are given without secutive years, the rate of change could no longer remain at the
further comment or warning to 9, 10 and more digits. At what point alleged error-free value.

do these figures lose their significance? Is it the same point in all Simple as these considerations are, they are basic. They illustrate
series, in a majority of series? No discussion, no commentary, precisely the importance of the just quoted statement from Gauss.
nothing: Just an uncritical abandonment to figures. To make matters worse, one is now no longer satisfied with yearly

Now we come to the point of looking at the calculations which comparisons, but uses quarterly data. For these the same
are made with the GNP figures. arrangements apply. In fact, the situation is worse, because in

It is well to start by quoting a statement made by one of the gathering the thousands of underlying figures, there are delays, cor-
greatest mathematicians of all time, C. F. W. Gauss: “The lack of rections, etc., in spite of the availability of electronic computers at
mathematical insight shows up in nothing as surprisingly as in un- certain stages. This is true whether yearly or quarterly figures are
bounded precision in numerical computations.” There are countless collected. The respective statistical offices naturally want to do as
many illustrations for such activity, not only in economics but in all good a job as possible and so they are led to correct the initially
social sciences. This is another sign for their lack of maturity com- given figures because faults are discovered over time. Indeed, the
pared to the physical sciences. There seems to be a love for yearly GNP, or even the much finer National Income Statistics,
meaningless digits. Norbert Wiener, also an eminent mathemati- often take up to ren years until the final figure is presented. In the
cian, upon looking at material described in my book, On the Ac- course of these successive revisions there are plus and minus changes.
curacy of Economic Observations (1963), observed to me: Expressed as percentages of the initially given figures the revisions
“Economics is a one-digit science.” I would be inclined to say that are of the order of 2, 3, or sometimes more percent. And over the
this is true for parts of it, but there are a few parts where two, years, some corrections go up, some go down for the same year.
perhaps even three, digits are admissible and where our power of in- Clearly, quarterly figures are subject to even more, and also bigger
terpretation is great enough to handle such fine detail. corrections. Now all this plays havoc with the initially calculated

The principal calculations to which GNP figures are subjected are rate of change, presumably of growth. Finally, even when after
rates of change, presumably rates of growth, because that is what the many years a definite figure is stated (different a/ways from the first
world has been made aware of. Governments everywhere look hyp- one!) that figure is not and cannot be free of some remaining error
notized at these calculations and are judged by the number of observations. Recall that no measurement is absolute, that we live

generated even for the short intervals of a quarter year — inciden- in a stochastic universe and that science — no matter how well
tally so brief a period that one must be astonished that (even given developed — is only an approximation to reality.
modern means of communications) all the hundreds of thousands of Now it is peculiar that governments are submitting to be judged

underlying figures could be collected: all this without any error by as primitive numbers as those here discussed. It is not that one
whatsoever! Or, if — heaven forbid — there are errors that they would want to deny that there exists some intuitive notion of


===== PAGE 6 =====

370 VITAL SPEECHES OF THE DAY

“growth,” that some parts of this phenomenon can be expressed interest rates, steep monopolistically controlled oil price increases are
numerically, but that all can be put into one single number that is with us. “Fine tuning” would require frequent switches in policy
free of all possible faults: that is unacceptable. What is more, these given the many factors which determine economic lite but such
numbers are supposed to be comparable over long periods of time switches are not only not welcome but can hardly be made, given the
and internationally. Whether it be India or Sweden or China or inherent slowness in political processes.

Japan or the U.S.: the precise rates of growth are compared with Summarizing: We would of course like to know whether and
each other. For some countries even the number of inhabitants is how much and in what direction the economy has grown and how
only vaguely known. It is difficult to see why statesmen would ex- economic welfare has been changed, presumably been improved.
pose themselves to this kind of evaluation. Economists have to share We would like to have good, trustworthy numerical expressions for
in the blame. They have introduced the notion of “fine tuning” of both. Alas, the GNP concept is primitive in the extreme and cer-
the economy, the idea that one could control the whole economy so tainly useless for any adequate expression of “growth.” “Social
precisely that its performance could be judged by the second digits of welfare" is still so difficult and controversial a notion that, without
growth rates, when in truth even the first digit is in doubt. One overcoming the inherent great conceptual difficulties we cannot in-
hears a little less, however, about the fine tuning recently in view of dicate any measure at present which is scientifically unchallengeable
the fact that rather gross events such as much unemployment, high and does not involve political, moral, or whatever other prejudices.

Freedom

A TIME TO CRY "WOLF"
By WILLARD C. BUTCHER, President, The Chase Manhattan Bank
Delivered before the Economic Club of Chicago, lllinois, February 20, 1975

'M very disturbed about what I see taking place in Itis estimated that it is fully two years older than that of Europe and

this country. 1 am disturbed about the economy — not Japan, and there is a fundamental correlation between modern plant

so much that we are in a recession as some of the senseless ways and productivity. A tabulation of growth rates by the OECD of
we try to get out of it. Not so much by inflation as by the fact we try twenty advanced economies for the 1960-1970 decade put the
to counteract and make up for it, rather than cure it. United States pretty close to the bottom — in eighteenth place. with

I am disturbed by the fact that people aren't being told enough average annual growth of only 4 percent. Japan headed the list with
about the essential nature of our economy and our economic an ll percent growth rate.

problems. I am distressed by the media, and by what seems to be It's significant that in this period Japan was putting about a third

either their failure to understand what makes our economy work, or of its GNP into investment spending, while we put less than a sixth

their inability or unwillingness to communicate that knowledge. of ours to work as capital. To catch up and to stay caught up can cost
I am disturbed by the kind of politics-as-usual in Washington as much as $225 billion over the next ten years.

that puts less emphasis on solving our problems than on who will get Our capital investment has not been enough because we fell

the credit if they are solved, or the blame if they are not. And I'm behind our needs in financing the search for more sources of energy.
disappointed in much of the nation’s business community, which is For energy self-sufficiency alone, it is estimated that we in this coun-
either too indifferent, too resigned, or too frightened to com- try will have to provide about $850 billion over the next ten years.

municate forcefully with the people. which equals about 80 percent of our total industrial investment in
I am troubled by the erosion of personal freedom that has taken the past ten years.
place and continues to take place, and by the complacency with It was not enough because we have fallen short of meeting our

which many Americans surrender their right to make choices, to transportation needs — for mass transportation of people and more
render judgments, and to exercise control over their own lives and energy-efficient transport of goods. which means mainly by rail.
destinies. That could require a ten-year investment of another $225 billion.
Fundamentally, I am concerned over the way we are giving up. Then there is the matter of our employment needs. There is no
bit by bit, what America really stands for. And before this evening is better — indeed no other — way to create new jobs than by supply-
over, I hope to have transmitted to you some of my sense of unease ing capital to provide the tools, the supplies, and the materials that
and alarm. jobs require. Economists tell us that it takes anywhere from $20,000
My subject today is capital formation and economic policy, which to $30,000 in capital investment to back up every worker in
takes in a lot of territory. So let's look at where we stand, where we American industry.
want to go, and what we need to get there. All in all, to meet these needs over the next ten years will require
To start with, our total private capital plant today amounts to more than twice as much capital as the last ten. How have we
some 3.2 trillion dollars, of which 1.8 trillion dollars represents our arrived at that figure? Considerable economic analysis indicates that

industrial capacity. Almost two-thirds of that amount — or 1.1 from today until early 1985, investment spending of 2.5 trillion con-
trillion — was developed and invested in just the past ten years. stant dollars is projected, on the assumption that we will see a con-

It was a tremendous achievement. But it was not enough. We did  tinuation of our relatively slow growth rate of 4 percent a year. Add
not meet some of our country’s basic needs. in inflation, at a presumed rate of approximately 5 percent a year,

It was not enough because our industrial plant today is deficient. and that comes to 3.6 trillion current dollars. If inflation were to be


===== PAGE 7 =====

Copyright © 2003 EBSCO Publishing