An Introduction to

Positive Economics
Sixth Edition

RICHARD G. LIPSEY
Sir Edward Peacock Professor of Economics,
Queen's University, Kingston, Ontario

EL BS
English Language Book Society/Weidenfeld & Nicolson
Weidenfeld & Nicolson Ltd
91 Clapham High Street, London SW4 9-1-1

© Richard G. Lipsey 1963, 1966, 1971, 19-5. 1979, 1983

All rights reserved. No part of this publication may be

reproduced, stored in a retrieval system, or transmitted in
any form or by any means, electronic, mechanical,
photocopying, recording or otherwise, without the prior
permission of the copyright owner.

First published 1963

Second edition 1966
Third edition 1971
Paperback edition first published 1972
Fourth edition 1975
Fifth edition 1979 Sixth
edition 1983
Reprinted 1985 (twice). 1986

ELBS edition first published 196:

Reprinted 1968, 1969 (twice) ELBS
edition of third edition 197 Reprinted
1973, 1974 (twice)
ELBS edition of fourth edition 1975
R e p r i n t e d 1 9 7 6 (t w i c e ) . 1 9 7 7
ELBS edition of fifth edition 1970
Reprinted 1980, 1981, 1982
ELBS edition of sixth edition 1983
Reprinted 1985 (twice). 1986

ISBN 0 297 78266 5

Printed in Great Britain by

Butler & Tanner Ltd,
Frome and London
To Diana
 Part one

Scope and method

Introduction

Why has the history of most capitalist countries been one of several years of boom and plenty
followed by several years of depression and unemployment with consequent poverty for
a great many citizens? Why, during the 1930s in most capitalist countries, was up to one
person in four unemployed while factories lay idle and raw materials went unused; why, in
short, was everything available to produce urgently needed goods and yet nothing happened?
Have we now learned, thanks to the theories propounded by many British and Continental
scholars and then synthesized and further developed by an English academic economist
writing over forty years ago from King's College, Cambridge, how to avoid for ever such
devastating (shocking) situations? Why, then, in the 1980s, did unemployment in Britain, the United
States and several other countries reach the highest levels ever attained since the Great Depression of
the 1930s? Is it really true that while the Great Depression of the 1930s overwhelmed unwitting and
powerless governments, the great slump (collapse) of the 1980s was deliberately (on purpose)
engineered by governments who knew full well what they were doing?
What determines exchange rates and why do they change? Why was one British pound sterling
worth 4.86 US dollars in 1930, 2.80 in 1950, 2.39 in 1970, 1.80 in 1976, 2.33 in 1980 and 1.65 at the
beginning of 1983? What is the point of international trade and would Britain be better off if it
reduced its dependence on trade? Are tariffs needed to protect home industries from unfair
competition, particularly from low-wage countries?
What determines the level of wages and what influences do unions have on the share of national
income going to labor? What functions do unions fulfill in today's world? Is it possible that having
fully achieved the purpose of putting labor on an equal footing with management they have
outlived (survived) their usefulness?
Must all modern economies make use of money? Could money be eliminated in a truly socialist
state? How it is that new money can be created by ordinary commercial banks within broad limits,
and by governments without limits? If money is valuable, why do economists insist that countries
with large supplies of it are no richer than countries with small supplies?
Why did inflation accelerate so dramatically in most countries in the mid-1970s? Why did
inflation peak at over 25 per cent in Britain and only 11 per cent in the United Slates? Are the group
of economists called monetarists right in arguing that the primary cause of British inflation is the
mismanagement of the British monetary system by that famous old institution, the Bank of
England?
What influence does government have on people's welfare? What are the effects of
3
4 Scope and method

government's taxation policies? What are the effects of public expenditure? How important to our
welfare is the size of our national debt?
These are the types of questions with which economists concern themselves, and on which the
theories of economics are designed to shed some light. Such a list may give you a better idea of the
scope of economics than could be obtained at this stage from an enumeration (listing) of the
common textbook definitions.
When you begin this book you are setting out on a study of positive economics or, to use a
slightly-more accurate phrase, POSITT ECONOMIC SCIENCE. From the questions listed above you
now have some idea of the scope of economics. Next we must consider in some detail what is meant by
the terms positive and science. After that we will ask whether or not it is really possible to conduct a
scientific study of anything that is basically concerned with human behavior. Economics is
generally regarded as a social science, but can any study of human behavior ever hope to be
`scientific'? Economists claim to be able to understand and to predict certain aspects of human
behavior. To anyone who wishes to be able to evaluate these claims this introductory discussion is
critical, because the questions 'What can we hope to learn?' and 'How can we go about it?' are basic to
the whole subject. These are also questions over which there is some disagreement among
professionals and a vast amount of misunderstanding and even superstition among the general
public.

Positive and normative statements

The success of modern science rests partly on the ability of scientists to separate their views on what
does happen from their views on what they would like to happen. For example, until the nineteenth
century virtually all Christians, Jews and Mohammedans believed that the earth was only a few
thousand years old. About two hundred years ago, evidence began to accumulate that some existing
rocks were millions of years old, possibly even thousands of millions. Most people found this hard
to accept; it would force them to rethink their religious beliefs and abandon those that were based
on a literal reading of the Bible. Many wanted the evidence to be wrong; they wanted rocks to be
only a few thousand years old. Nevertheless, the evidence continued to accumulate until today
virtually everyone accepts that the earth is neither thousands, nor millions, but four or five thousand
million years old. This advance in our knowledge came because the question 'H ow old are
observable rocks?' could be separated from the feelings of scientists (many of them devoutly
religious) about the age they would have liked the rocks to be.

Definitions and illustrations

Distinguishing what is from what we would like, or what we feel ought, to be.' depends partly on
knowing the difference between positive and normative statements.

' This word 'ought' has two distinct meanings: the 'logical ought' and the 'ethical ought'. The logical ought refers to the
consequences of certain things: e.g., 'you ought to leave now if you don't want to be late'. The ethical ought refers to the
desirability of certain things: e.g., `you ought to leave now because it is impolite to stay too long'. The text obviously refers to the
ethical ought.
 Introduction 5

Positive statements are about what is, was or will be: they assert alleged (assumed) facts
about the universe in which we live. Normative statements are about what ought to
be. They depend on our judgments about what is good or bad, and they are thus
inextricably (impossible to separate them) bound up with our philosophical, cultural and
religious positions.
We say that normative statements depend on our VALUE JUDGEMENTS.
Let us consider some assertions (claim), questions and hypotheses that can be classified as positive
or normative. The statement ‘it is impossible to break up atoms' is positive one that can
quite definitely be (and of course has been) refuted (prove false) by empirical observations,
while the statement 'Scientists ought not to break up atoms' is a normative statement that involves
ethical judgments, and cannot be proved right or wrong by any amount of evidence. In economics the
questions 'What policies will reduce unemployment?' and 'What policies will prevent inflation?' are
positive ones, while the question 'Ought we to be more concerned about unemployment than about
inflation?' is a normative one. The statement 'An increase in government spending will reduce
unemployment and increase inflation' is a positive hypothesis. The statement 'Unemployment is a more
serious social problem than inflation' is normative.

As an example of the importance of this distinction in the social sciences, consider the question
`Has the payment of generous unemployment benefits increased the amount of unemployment?'
This positive question can be turned into a testable hypothesis by asserting something like: 'The
higher are the benefits paid to the unemployed, the higher will be the total amount of unemployment.’ If
we are not careful, however, our attitudes and value judgments may get in the way of our study of
this hypothesis. Some people are opposed to the welfare state and believe in an individualist, self-help
ethic. They may hope that the hypothesis will be found correct because its truth could then be used as
an argument against welfare measures in general. Others feel that the welfare state is a good thing,
reducing misery and contributing to human dignity (self-respect). They may hope that the hypo-
thesis is wrong because they do not want welfare measures to produce results of which people
disapprove. In spite of different value judgments and social attitudes, however, evidence is
accumulating on this particular hypothesis. As a result, we have much more knowledge than we had
ten years ago of why, where and by how much (if at all) unemployment benefits increase
unemployment. This evidence could never have been accumulated or accepted if investigators had
not been able to distinguish their feelings on how they wanted the answer to turn out from their
assessment of evidence on how people actually behaved.

Positive statements such as the one just considered assert things about the world. If it is possible
for a statement to be proved wrong by empirical evidence, we call it a TESTABLE STATEMENT.
Many positive statements are testable, and disagreements over them are appropriately handled
by an appeal to the facts.

In contrast to positive statements, which are often testable, normative statements are never
testable. Disagreements over such normative statements as 'It is wrong to steal' or 'It is immoral to
have sexual relations out of wedlock’ cannot be settled by an appeal to empirical observations.
Thus, for a rational consideration of normative questions, different techniques are needed to those
used for a rational consideration of positive questions. Because of this, it is convenient to separate
normative and positive enquiries. We do this not because we think the former are less important
tri dignity an the latter, but merely because they must be handled in different ways.
6 Scope and method

Some points of possible confusion

Having made this distinction between positive and normative, a number of related points require
mention. Although we deal with them only briefly, any one of them could be the subject of extended
discussion.
The classification is not exhaustive (complete). The classifications 'positive' and 'normative' do not cover
all statements that can be made. For example, there is an important class, called analytic statements,
whose truth or falsehood depends only on the rules of logic. Consider the single sentence: `If every X
has the characteristic Y, and if this item Z is in fact an X, then it has the characteristic Y.' This
sentence is true by the rules of logic, and its truth is independent of what particular items we
substitute for X, Y, and Z. Thus the sentence ‘all men are immortal and if you are a man, then you are
immortal' is a true analytic statement. It tells us that if two things are true then a third thing must be
true. The truth of this statement is not dependent on whether or not its individual parts are in fact true.
Indeed the sentence 'All men are immortal' is a positive statement which has been refuted by myriad
(many) deaths. Yet no amount of empirical evidence on the mortality of men can upset the truth of
the sentence ‘all men are immortal and if you are a man, then you are immortal.'

Not all positive statements are testable A positive statement asserts something about the
universe. It may be empirically true or false in the sense that what it asserts may or may not be true of
the universe. If it is true, it adds to our knowledge of what can and cannot happen. Many positive
statements are refutable: if they are wrong this can be ascertained (determined) (within a margin
for error of observation) by checking them against data. For example, the positive statement that
the earth is less than five thousand years old was tested and refuted by a mass of evidence which had
been accumulated in the nineteenth century. The statement 'Angels exist and frequently visit the earth in
visible form' is, however, also a positive statement. It asserts something about the universe. But we
could never refute this statement with evidence because, no matter how hard we searched, believers
could argue that we did not look in the right places or in the right way, or that angels won't reveal
themselves to non-believers, or any one of a host of other alibis (explanation). Thus
statements that could conceivably be refuted by evidence if they are wrong are a subclass of
positive statements: other positive statements are irrefutable.

The distinction is not unerringly (accurately) applied because the positive-normative

distinction helps the advancement of knowledge, it does not follow that all scientists automatically
and unerringly apply it. Scientists are human beings. Many have strongly held values and they
may let their value judgments get in the way of their assessment of evidence. For example, many
scientists are not even prepared to consider evidence that there may be differences in intelligence
among races because as good liberals they feel that all races ought to be equal. Nonetheless, the
desire to separate what is from what we would like to be is a guiding light, an ideal, of science. The
ability to do so, albeit imperfectly, is attested (confirm) to by the acceptance, first by scientists and
then by the general public. of many ideas that were initially extremely unpalatable (bad) - ideas such
as the extreme age of the earth and the evolution of man from other animal species.

Ideals can be important even though they are not universally applied. Consider an analogy. (1) Many
people try to be good (according to their own lights). (2) Most people do not live up to their own
standards of goodness all of the time. (3) Ideas of goodness are an important force in motivating
 Introduction 7

human behavior. All three of these statements are probably true: the truth of (1) does not preclude
the truth of (2) and the truth of (2) does not preclude the truth of (3). In an analogous way all three of
the following statements might be true. (1) Positive and normative statements can be distinguished. (2)
Not all scientists do, or even could, maintain the distinction all of the time. (3) The distinction has
been a potent (strong) force in the advancement of knowledge and in the separation of
knowledge from prejudice. Statement (1) does not preclude (2) and (2) does not preclude (3).
Economists do not need to confine their discussions merely to positive statements. Some
critics have mistakenly assumed that economists must try to deal only in statements that are positive and
testable. In fact the positive economist must spend time worrying about the correctness of
analytic statements: 'Is a certain prediction actually implied by a certain set of assumptions?' He
must also be prepared to have any number of non-testable assumptions in his theory as long as some
testable predictions can be deduced from it. Also, he should not shrink from discussing value
judgments, as long as he knows what he is doing. From the fact that positive economics does not
include normative questions (because its tools are inappropriate to them) it does not follow that the
student of positive economics must stop his inquiry as soon as someone says the word ought. The
pursuit of what appears to be a normative statement will often turn up positive hypotheses on which our
ought conclusion depends. For example, although many people wax quite emotional for or
against government control of industry, probably few of them believe that government control is
good or bad in itself. Their advocacy or opposition will be based on certain beliefs about relations
which can be stated as positive rather than normative hypotheses. For example: 'Government
control reduces (increases) efficiency, changes (does not change) the distribution of income, leads
(does not lead) to an increase of state control in other spheres.' A careful study of this emotive
subject will reveal an agenda for positive economic inquiry that could keep a research team of
economists occupied for a decade.
The nature of positive economics
Positive economics is concerned with the development of knowledge about the behavior of people and
things in the world. This means that its practitioners are concerned with developing propositions
that fall into the positive, testable class. This does not mean, however, that every single statement and
hypothesis to be found in positive economics will actually be positive and testable. Some time ago a
philosophy of knowledge called logical positivism was popular. It held that every single statement in
the theory had to be positive and testable. This proved to be a harmful and unnecessary
straitjacket.
All that the positive economist asks is that something that is positive and testable should
emerge from his theories somewhere – for if it does not, his theories will have no relation to
the world around him.

1
Many critics of the idea of positive science have argued otherwise. They feel that because no person can ever be perfectly
k
-,P o bjective about other people, the idea of an objective, fact-guided science of human behaviour is a contradiction. .
- ,Ortunately, science based on the testingg of positive hypotheses is possible even though no one individual can be relied on
c
ompletely and always to scparee his jud ement of facts from his desires on what he would like the facts to be.
8 Scope and method

The positive economist seeks ways of answering positive testable questions such - lose listed at the
outset of this chapter. His approach to these questions can, in a general way, be described as
scientific. We must now consider in more detail just what the scientific approach is, and how
scientific theories are developed and used.

The scientific approach

Very roughly speaking, the scientific approach consists in relating questions to evidence. This
approach to a problem is what sets scientific inquiries off from other kinds of inquiries.'
In some fields, the scientist is able to generate observations that will provide evidence concerning
any hypothesis that he wishes to test. Experimental sciences, such as chemistry and some branches of
psychology, have an advantage because it is possible to produce relevant evidence through
controlled laboratory experiments. Other sciences, such as astronomy and economics, cannot do
this. They must wait for time to throw up observations that may be used to test hypotheses.
The ease or difficulty with which one can collect evidence does not determine whether or not a
subject is scientific, although many people have thought otherwise.' The procedure of scientific
inquiry does, however, differ radically between fields in which laboratory experiment is possible and
those in which it is not. Here we consider general problems more or less common to all sciences. In
Chapter 3 we deal with problems peculiar to the non-experimental sciences, which must accept those
observations that the world of actual experience provides.
It is often said that we live in a scientific age. Over the last several hundred years the citizens of most
Western countries have enjoyed the fruits of innumerable (countless) scientific discoveries. But
the scientific advances that have so profoundly affected the average citizen have been made by
an extremely small minority of the population. Most people have accepted these advances without the
slightest idea either of the technical nature of the discoveries involved, or of the attitude of mind that
made them possible. If we take as a measure of the influence of science the degree of dissemination
(distribution) of the fruits of science, then we live in a profoundly scientific age; but if we take as
our measure the degree to which the general public understands and practices the scientific approach,
then we are definitely in a pre-scientific era. Indeed, the scientific method of answering questions by
appealing to a carefully collected and coordinated body of facts is a method that is seldom
adopted by the public.

Consider, for example, the argument about capital punishment that continues even in many of
the countries that have abolished (eliminate) the death penalty. It is possible to advocate capital
punishment as an act of pure vengeance, or because we believe that morally a person who kills ought
himself to be killed. If we argue about capital punishment on these grounds, we are involved in
normative questions depending upon value judgments. The great majority of arguments for capital -

Other approaches might be to appeal to authority. for example. to Aristotle or the Scriptures, to appeal by introspection to
some inner experience (to start off 'all reasonable men will surely Ligree'). ur to proceed by way of definii_ic.ins to the 'true'
nature of the problem or concepts under consideration.
2 It is often thought that scientific procedure consists of grinding out ans \ikers by following blind rules of calculation, and

that only in the arts is exercise of real imagination required. This view is misguided. What the scientific method gives is an
impersonal set of criteria for answering some questions. What questions to ask, exactly how to ask them and how to obtain
the evidence are difficult problems for which there are no rules. They require, upon occasion, great feats of imagination and
ingenuity.
 Introduction 9

Punishment, however, are not of this type. Instead, they depend on predictions about
observable behavior, and thus belong to the field of science. These are usually variants of the general
argument that capital punishment is a deterrent to murder.
It is truly amazing how people can become committed to agreeing or disagreeing with this
proposition without considering the available evidence. How many people involved in these debates
know anything, for example, of the mass evidence on murder rates before and after the abolition of
capital punishment in the large number of jurisdictions where it has been abolished and in those few
where it has been reimposed? Indeed most popular arguments for and against capital punishment
involve a maximum of empirical questions and a minimum of empirical evidence used to arrive at
the answers given. if we really believed in a scientific inquiry into human behavior, we would try to
state the arguments about capital punishment in terms of a specific set of propositions and then set out
systematically to gather evidence relating to each one.
We may conclude that many hotly debated issues of public policy involve positive not norma-
tive, questions, but that the scientific approach to them is very often avoided.

-
A science of human behaviour?
thee preceding discussion raises the question of whether or not it is possible to have a scientific study the
field of human behaviour i n v a r i o u s k i n d s o f s c i e n c e s
It is often argued that natural sciences deal with inanimate matter that is subject to natural 'laws',
while the social sciences deal with man, who has free will and cannot, therefore, be made the subject
of . such (inexorable) laws. Such an argument, however, concentrates on the physical sciences; it
/nits biology and the other life sciences which deal successfully with animate matter. When this
ant is granted, it may then be argued that the life sciences deal with simple living material, while
77
Cittly the social sciences deal with human beings who are the ultimate in complexity and who alone
ss free will. Today, when we are increasingly aware of our common heritage with apes (chimpanzee) in
Particular, and primates (monkeys) in general, an argument that man's behavior is totally different from the
behaviour of other animals finds few adherents among informed students of animal behavior.

Human behaviour

Nonetheless, many social observers, while accepting the success of the natural and the life sciences, hold
that there cannot be a successful social science. Stated carefully, th is v ice implies that
inanimate (lifeless) and non-human animate matter will show stable responses to certain
stimuli, while humans will not. For example, if you put a match to a dry piece of paper the paper will burn,
while if you try to extract information from unwilling human beings by torture, some will yield it and
others will not. More confusingly, the same individual reacts differently at different times.
Whether human behavio r does or does not show sufficiently stable responses to factors influencing it as
to be predictable within an acceptable margin of error is a positive question that can only be --settled by
an appeal to evidence and not by a priori speculation.'
itriori is a phrase commonly used by economists. It may be defined as that which is prior to actual experience, or as that
Is innate or based on innate ideas.
iisakott.;-
10 Scope and method

In fact, it is a matter of simple observation that when we consider a group of individuals they do
not behave in a totally capricious (unpredictable) way but do display stable responses to various
stimuli. The warmer the weather, for example, the higher the number of people visiting the
beaches and the higher the sales of ice-cream. It may be hard to say when or why one individual
will buy an ice-cream, but we can observe a stable response pattern from a large group of individuals:
the higher the temperature the greater the sales of ice-cream.
Many other examples will come to mind where, because we can say what the individual will
probably do – without being certain of what he will do – we can say with quite remarkable accuracy
what a large group of individuals will do. No social scientist could predict, for example, when an
apparently healthy individual is going to die, but death rates for large groups are stable enough to
make life insurance a profitable business. It could not be so if group behavior were capricious.
Also, no social scientist can predict what particular individuals will be killed in car accidents next
holiday, but we can come very close to knowing how many in total and the more objectively
measurable data we have concerning, for example, the state of the weather on the day, and the
increase in car sales over the last year, the closer we will be able to predict the total of deaths.

If group human behavior were in fact random and capricious (unpredictable), existence would be
impossible. Neither law, nor justice, nor airline timetables would be more reliable than a roulette wheel;
a kind remark could as easily provoke fury as sympathy; one's landlady might put one out
tomorrow or forgive one the rent. One cannot really imagine a society of human beings that could
possibly work like this. Indeed a major part of brainwashing techniques is to mix up rewards and
punishments until the victim genuinely does not know 'where he is': unpredictable pressures drive
human beings mad. In fact, we live in a world which is some sort of mixture of the predictable, or
average, or 'most of the people most of the time.' and of the haphazard, contrary, and random.

When we try to analyze our world, and apply our orderly models to it, we need help from
specialists in probability — statisticians – but we have not yet found that we need the
advice of experts in the behavior of systems in states of total chaos (confusion).

The 'law' of large numbers

We may now ask how it is that we can predict group behavior when we are never certain what a
single individual will do. As a first step, we must distinguish between deterministic and statistical
hypotheses. Deterministic hypotheses admit of no exceptions. An example would be the statement: 'If
you torture any man over this period of time with these methods he will always break down.'
Statistical hypotheses, however, admit of exceptions and purport (claim) to predict the
probability of certain occurrences. An example would he: 'if you torture a man over this period of
time with these methods he will very probably break down: in fact if you torture a large number of
men under the stated circumstances about 95 per cent of them will break down.' Such a
hypothesis does not predict what an individual will certainly do, but only what he will probably do.
This does allow us, however, to predict within a determinable margin of error what a large group of
individuals will do.

Successful predictions about the behavior of large groups are made possible by the statistical
'law' of large numbers. Very roughly, this 'law' asserts that random movements of a large number of
items tend to offset one another. The law is based on one of the most beautiful constants of
behavior in the whole of science, and yet the law can be derived from the fact that human beings make errors!
This constant is the normal curve of error which you will encounter in elementary statistics.
 Introduction 11

Let us consider what is implied by the law of large numbers. Ask one person to measure the
length of a room and it will be almost impossible to predict in advance what sort of error of
measurement he will make. Thousands of things will affect the accuracy of his measurements.
Furthermore, he may make one error today and quite a different one tomorrow. But ask one
thousand people to measure the length of the same room and we can predict with a high degree of
accuracy how this group will make its errors! We can assert with confidence that more people will
make small errors than will make large errors, that the larger the error the fewer will be the number
of people making it, that the same number of people will overestimate as wil1 underestimate the
distance, and that the average error of all the individuals will be zero.' Here then is a truly
remarkable constant pattern of human behavior; a constant on which much of the theory of
statistical inference is based.

If a common cause should act on all members of the group we can successfully predict their
average behavior, even though any one member of the group may act in a surprising fashion. If, for
example, we give all our thousand individuals a tape measure which understates 'actual' distances,
we can predict that, on the average, the group will now understate the length of the room. It is, of
course, quite possible that one member who had in the past been consistently under measuring
distance because he was depressed will now overestimate the distance because the state of his health
has changed; but something else may happen to some other individual that will turn him from an over
measurer into an under measurer. Individuals may do peculiar (strange) things which, as far as we can
see, are inexplicable, but the group's behavior, when the inaccurate tape measure is substituted for the
accurate one, will nonetheless be predictable, precisely because the odd things that one individual does tend
to cancel out the odd things some other individual does.

The nature of scientific theories

So far we have seen that there is real evidence that human behavior does show stable response
patterns. Theories grow up in answer to the question 'Why?' Some sequence of events, some
regularity between two or more things is observed in the real world and someone asks why this
should be so. A theory attempts to explain why. One of the main practical consequences of a theory is
that it enables us to predict as yet unobserved events. For example, national income theory
predicts that a cut in tax rates will reduce the amount of unemployment. The simple theory of
market behavior predicts that if there is a partial failure of the potato crop, the total income
earned by potato farmers will increase!

The construction of theories

A theory consists of a set of definitions, stating clearly what is meant by various terms, and a set of
'Forpurposes of measuring the error we define the `true' distance to be that measured by the most precise instruments of
scientific measurement (whose range of error will be very small relative to the range or error of our one thousand laymen all
Wielding tape measures). Those familiar Aith statistical theory will realize that the predictions in toe text assume that all the

accessary conditions, such as the existence of a large number of independent factors causing individuals to make errors, are

.E.:Juifilled- The purpose of the discussion in the text is not to give readers a full appreciation of the subtleties otstatistical
y, but to persuade them that anyone is misguided who holds the common view that free will and the absence of
erministic certainty about human behaviour makes a scientific study of such behaviour impossible.
det
1 2 Scope and method

assumptions about the way in which the world behaves. Students often worry about what appears to
them to be the unrealistic assumptions that they encounter in economics. It is important to remember,
however, that all theory is an abstraction from reality. If we did not abstract we would merely
duplicate the world camera-style, and would add nothing to our understanding of it.

A good theory abstracts from reality in a useful way; a bad theory does not.

But how do we know if the abstractions of a particular theory are useful or not? To do this we
must take the next step in the construction of theories: we follow a process of logical deduction to
discover what is implied by the assumptions of the theory. For example, if we assume that businessmen
always try to make as much profit as is possible, and if we make assumptions about how taxes affect
their profits, we can derive implications about how they will behave when taxes change. These
implications are the predictions of our theory. If the theory is useful, its predictions will pass
empirical tests. For example, when taxes are next changed, businessmen will react in the way
predicted by the theory.

Scientific predictions

We have seen above that a successful theory enables us to predict events. What is the nature of a
scientific prediction, and is it the same thing as a prophecy about the future course of events?

A scientific prediction is a conditional statement having the form 'if you do this
then such and such will follow.'
If you mix hydrogen and oxygen under specified conditions, then water will be the result. If the
government has a large budget deficit, then unemployment will be reduced. It is most important to
notice that this prediction is very different from the statement: 'I prophesy that in two years' time
there will be a large increase in employment because I believe the government will decide to have a
large budget deficit.' The government's decision to have a budget deficit or surplus in two years' time
will be the outcome of many complex factors, emotions, objective circumstances and chance
occurrences most of which cannot be predicted. If the economist's prophecy about the level of
employment turns out to be wrong, because in two years' time the government does not have a large
deficit, then all we have learned is that the economist is not a good guesser about the behavior of the
government; we will not have found evidence that conflicts with any economic theory. However, if
the government does have a large deficit (in two years' time or at any other time) and then
unemployment does not fall, we have found evidence conflicting with a (conditional) scientific
predict in the field of economics.

It p r y important not to treat economic forecasting as synonymous with economic prediction. Forecas:.ng is a7ype of

condition.:: prediction which attempts to predict the future by discovering relations between economic variables of 7he sort
that the LIC of Y at some future date depends on the value °IX toda„,,, in which case future lean be predicte by ob,erving
present X. _Many conditional predictions are not of this form; those which relate the Ytoday to the value of today provide
significant and useful relations that allow us to predict 'if you do this to X you will do that to Y', without allowing us to
forecast the future. The analogy often drawn between economics and weather forecasting relates to economic forecasting
rather than to the wider class of scientific economic predictions.
 Introduction 13

The testing of scientific predictions

If we wish to test any theory we confront its predictions with evidence. We seek to discover if certain
events have the consequences predicted by the theory. Some of the difficult problems involved in this
task are the subject of Chapter 3. In the meantime we should notice that as with most other sciences it
is never possible to prove e or to refute any theory in economics with 100 per cent certainty.

Proof and refutation Consider the simple economic theory that predicts: 'If a sales tax is levied on
the product of a competitive industry, then the price of the product will rise but by less than the
amount of the tax.' It is not claimed that this prediction holds only for the years 1970-95, or only in
Odd-numbered years, nor is it supposed to hold in the USA and Germany but not in France and
Paraguay. The theory simply says that this result will hold whenever a sales tax is levied in an industry
that is competitive. We may say that the theory is unbounded both in time and in space. But since we can
make only a limited number of observations, we can never prove conclusively that the theory is true.
Even if we have made a thousand observations which agree with the prediction, it is always possible
that in the future we will make observations which conflict with the theory. Since this posibility
can never be ruled out completely (no matter how unlikely we might think it to be), we never can
regard any theory as conclusively proved.
It is also impossible to refute any theory conclusively. This matter is considered in some detail in pter
3. Suffice it to say now that, since human beings make the tests, and since human beings are fallible,
it is always possible that a piece of apparently conflicting evidence arose because we made a mistake
in our observations. One conflicting observation does not worry us very much, but as a mass of
them accumulate we become more and more worried about our theory and will regard it as less and
less likely to be true. Eventually we shall abandon it. even though we can never be 100 per ';;;Cent
certain we are not making an error in doing so.

When is a theory abandoned? As a generalization we can say that our theories tend to be
abandoned when they are no longer useful, and that they cease to be useful when they cannot predict the
consequences of actions in which we are interested better than the next best alternative. When this
happens the theory is abandoned and replaced by the superior alternative. In the process of
upsetting existing theories we learn new, surprising facts.
Any developing science will continually be having some of its theories rejected; it will also be
cataloguing (classification) observations that cannot be fitted into (explained by) any existing
theory. These observations indicate the direction required for the development of new theories or for
the extension of existing ones.' On the other hand, there will be many implications of existing theories
that have not yet been tested, either because no one has yet figured out how to test them, or merely
because no one has got around to testing them. These untested hypotheses provide agenda for new
empirical studies.

I The development of a nev+ theory to account for ex sting observations is often the result of real creative genius of an
i
almost inspired nature. This step in the development of science is the exact opposite of the popular conception of the scientist as
an automatic rule-follower. One could argue for a long time whether there was more original creative genius embodied in a first-
class symphony or a new theory of astronomy. Fascinating studies of the creative process may be found in A. Koestler,
gPie Sleep Walkers (Hutchinson, 1959), especially the section on Kepler, and J. D. Watson, The Double Helix (Wiedenfeld and
Nicolson, 1968).
14 Scope and method

The state of economics

Economics provides no exception to the comments made in the previous paragraph. On the one
hand, there are many observations for which no fully satisfactory theoretical explanation exists. On the
other hand, there are many predictions which no one has yet satisfactorily tested. Thus serious
students of economics must not expect to find a set of answers to all possible questions as they
progress in their study. They must expect very often to encounter nothing more than a set of
problems for further theoretical or empirical research. Even when they do find answers to problems, they
should accept these answers as tentative and ask, even of the most time-honoured theory: `What
observations might we make that would be in conflict with this theory?' Economics is still a very
young science and many problems in it are almost untouched. Those of you who venture further
in this book may well, only a few years from now, publish a theory to account for some of the problems
mentioned herein, or else you may make a set of observations which will upset some time-honoured
theory described within these pages.

Having counselled disrespect for the authority of accepted theory, it is necessary to warn against
adopting an approach that is too cavalier. No respect attaches to the person who merely says: `This
theory is for the birds; it is obviously wrong.' This is too cheap. To criticize a theory on logical
grounds (economists sometimes say 'on theoretical grounds'), one must show that its alleged
predictions do not follow from its assumptions. To criticize a theory effectively on empirical
grounds, one must demonstrate by a carefully made set of observations that some aspect of the
theory is contradicted by the facts. These tasks are seldom easily or lightly accomplished.
Figure t_l provides a summary of the discussion of theories. It shows a closed circuit, because
theory and observation are in continuous interaction with each other. Starting at the top left
(because we must start somewhere), we find the definitions of terms and assumptions of a theory.
The theorist then deduces by logical analysis everything that is implied by the assumptions. These
implications are the predictions of the theory. The theory is then tested by confronting its predic-
tions with evidence. Tests may be made for the direct purpose of testing a theory, but they are also
made incidentally whenever an applied economist uses the theory in a real-world context. If the
theory is in conflict with facts, it will most likely be amended to make it consistent with the new facts
(and thus make it a better theory); in extreme cases it will be discarded in place of a superior
alternative. The process then begins again as the new or amended theory is subjected first to logical
analysis and then to empirical testing.

Scientific crises
Sciences often appear to evolve through a series of stages. At first, an existing theory seems to be
working well and the main scientific tasks are to extend it in various directions, Then, gradually.
observations begin to accumulate that conflict with the theory. For a long time these exceptions are
explained away on an ad hoc basis, but eventually the weight of conflicting evidence causes a crisis
for the theory. Finally a breakthrough occurs, and some genius develops a new theory that com-
prehends both what still seems right in the older theory and the observations that were not accounted
for. Once the new theory is accepted, often after an interlude of uncertainty and heated
controversy, another period of consolidation and extension occurs until new conflicd between
theory and observation emerge.
 introduction 15

Definitions and assumptions about

Logical deduction
(theoretical

Predictions
ti.',(often called implications)

The theory
is amended ii
light of the newly acc,ired facts

• Observation and
statistical analysis of the data The theory is
discarded in favour
of a superior
competing
theory

Iusion: the theory does or d o e s n o t provide either or

better explanation of the facts than alternative
theories If the theory is in conflict
with

If the theory passes the

test no consequent action is made
necessary by the test, although we
should
subject any theory

e interaction of deduction and measurement in theorizing

 16 Scope and method

Economics appears to be currently in one of these times of crises. After a long period of
consolidation and development of the theories of 'macroeconomics' (see Part 7), evidence has
accumulated that is not easily explained within the framework of existing theories. As is usual in
such times, there is substantial confusion and controversy. Some economists hold that only minor
amendments are needed to make existing theories consistent with the new observations; others
believe that major theoretical upheavals are necessary. In the heat of these controversies, econo-
mists tend to reveal what most scientists reveal in such situations: that they are not dispassionate
calculating machines, but human beings with strong emotional commitments to their established
positions.
Periods of scientific crisis can be profoundly disturbing for the scientists who become involved in
them, to say nothing of those who depend on the scientists for answers to practical questions. Many
economists are so committed to particular theories that they will never be convinced by new
evidence. It is important, however, that one of the rules of debate should continue to be 'Try to show
that your theory fits the evidence better than do competing theories'. Although the most committed
protagonists may never change their minds, a new generation of economists, not so committed to
old and outdated positions, may be able to judge the issues more dispassionately and be able to tell
which of various competing theories conforms more closely with the evidence.

Science has been successful in spite of the fact that individual scientists have not always
been totally objective. Individuals may passionately resist the apparent implications of
evidence. But the rule of the game, that facts cannot be Ignored and must somehow be
fitted into the accepted theoretical structure, tends to produce scientific advance in spite of
what might be thought of as unscientific, emotional attitudes on the part of many
scientists, particularly at times of scientific crisis.

But if existing protagonists ever succeed in changing the rules of the game by encouraging economists
to ignore inconvenient facts or define them out of existence, this would be a major blow to scientific
enquiry in economics.'

' One of the best introductions to methodology for economists is Mark Blaug, The Methodology of Positive Economics:
Or How Economists Explain (Cambridge University Press, 1980), which is, however, probably better read alter one has
studied a certain amount of economics.