Foundations of Economic Method

A Popperian Perspective, 2nd Edition

Lawrence A. Boland

• Publisher: Routledge
• Year: 2020
• ISBN:
  9781000159332
• Edition: 1
• Topic: Economics
• Subtopic: Economic History
• Index: Economics

Many consider Foundations of Economic Method to be Lawrence Boland's best work. This updated edition is radically changed from the original and will be much appreciated by thinkers within economics. The book positions methodology vis-à-vis the current practice of economists and is all the better for it. Yet another book that not only deserves to be read by those within the field of economic methodology, but also by those involved in economics at all. Boland is back.

Part I

THE ‘HIDDEN AGENDA’ OF NEOCLASSICAL ECONOMICS

1 The Problem of Induction vs. the Problem with Induction

Hume’s objective doctrine … has two parts: … (1) in causation there is no indefinable relation except conjunction or succession; (2) induction by simple enumeration is not a valid form of argument. Empiricists in general have accepted the first of these theses and rejected the second. When I say they have rejected the second, I mean that they have believed that, given a sufficiently vast accumulation of instances of a conjunction, the likelihood of the conjunction being found in the next instance will exceed a half; or, if they have not held exactly this, they have maintained some doctrine having similar consequences.

Bertrand Russell [1945, p. 667]

Hume showed that inductive arguments could not be justified, even in part, but he did not think that they were thereby incorrect. Most later writers have agreed.

David Miller [1994, p. 13]

Scientists never ‘explain’ any behavior, by theory or by any other hook. Every description that is superseded by a ‘deeper explanation’ turns out upon careful examination to have been replaced by still another description, albeit possibly a more useful description that covers and illuminates a wider area. I can illustrate by what everyone will agree is the single most successful ‘theory’ of all time. I refer to Newton’s theory of universal gravitation.

Paul Samuelson [1964, p. 737]

Since the time when Adam Smith’s friend David Hume observed that there was no logical justification for the common belief that much of our empirical knowledge was based on inductive proofs [Hume 1739; Russell 1945], methodologists and philosophers have been plagued with what they call the ‘Problem of Induction’. The paradigmatic instance of the Problem of Induction is the realization that we cannot provide an inductive proof that ‘the sun will rise tomorrow’. This leads many of us to ask, ‘So how do we know that the sun will rise tomorrow?’ If it is impossible to provide a proof, then presumably we would have to admit that we do not know! Several writers have claimed to have solved this famous problem [Popper 1972; Hollis and Nell 1975; see Miller 2002] – which is quite surprising, since it is impossible to solve. Nevertheless, what it is and how it is either ‘solved’ or circumvented is fundamental to understanding all contemporary methodological discussions.

The Problem of Induction

It is clear, Hume felt, that sense experience is the primary matter of all knowledge; ideas, general concepts, theories, universals, and all such things are secondary or derivative. This contention, that all knowledge is derivative from sense experience, leads directly to the ‘problem of induction’. No matter how many swans I have seen, nor how many have been seen by others, there is no justification for asserting the general proposition that ‘all swans are white’.

Scott Gordon [1991, p. 127]

Since the Problem of Induction is fundamental, a clear statement of it is needed. Before attempting this, let me clarify some of its elementary parts. First, there is the implicit presumption that empirical knowledge requires logical justification. I will call this ‘Justificationism’. Justificationism probably needs little explanation at this stage, since it is widely presumed or accepted, but for future reference, let me be specific.

Justificationism is the methodological doctrine that asserts that nobody can claim to possess knowledge unless he or she can also demonstrate (with a proof) that his or her knowledge is true; that is, everyone must justify his or her knowledge claims.

Crudely stated, this requirement says, ‘knowledge’ is not Knowledge unless it is (proven) true knowledge. Second, there is the further requirement that the justification of empirical knowledge requires an inductive, as opposed to a deductive, proof. This additional requirement will be called ‘Inductivism’. Although Inductivism has been around for several hundred years, the operative view of it will be the following:

Inductivism is the methodological doctrine that asserts that any justification of one’s knowledge must be logically based only on experiential evidence consisting of particular or singular observation statements; that is, one must justify his or her knowledge using only verifiable observations that have been verified by experience.

Given Inductivism, any straightforward solution to the Problem of Induction requires an ‘Inductive logic’, that is, there must be a form of logic which permits arguments consisting of only ‘singular statements’ (e.g., ‘The sun rose in Vancouver at 7:03am on November the 2nd, 2002’), while the conclusions that validly follow may be ‘general statements’ (e.g., ‘The sun will rise every day’). Now I can state the famous problem:

The Problem of Induction is that of finding a general method of providing an inductive proof for anyone’s claim to empirical knowledge.

In other words, this is the problem of finding a form of logical argument in which (a) the conclusion is a general statement, such as one of the true ‘laws’ of economics, or the conclusion is the choice of the true theory or model from among various competitors; and (b) the assumptions include only singular statements of particulars (such as simple observation reports). With an argument of this form one is said to be arguing inductively from the truth of particulars to the truth of generals. (In contrast, a deductive form of argument supposedly proceeds from the truth of generals to the truth of particulars.) If one could solve the Problem of Induction, the true ‘laws’ or general theories of economics (i.e., economic knowledge) could then be said to be induced logically from particular observations (and thereby justified).

For very many, many years virtually everyone believed that science and its ‘scientific method’ represented a solution to the Problem of Induction [see Agassi 1963]. This belief was based on the commonly accepted view that Newtonian physics represented true knowledge, since there were many reports of the existence of inductive proofs of that knowledge. Late in the nineteenth century, when doubts were raised concerning the absolute truth of Newtonian physics, a more moderate claim for science was developed [e.g., Poincaré 1905/52; Duhem 1906/62; Eddington 1928].

The Problem of Induction in economics

All theory depends on assumptions which are not quite true. That is what makes it theory.

Robert Solow [1956, 65]

It is interesting to note that except for some earlier books explicitly about methodology [e.g., Hollis and Nell 1975; Stewart 1979; Blaug 1980/92], economics writers have rarely been concerned with this allegedly fundamental problem. There is a very simple reason for this. For most of the nineteenth century, economists simply believed that the Problem of Induction had been solved; thus it did not need any further consideration. After all, Newton seems to claim to have arrived at the laws of physics from scientific observation using inductive methods [e.g., Newton 1704/1952]. In Adam Smith’s time, inductive generalization was the paradigm of rational thinking; Newton’s physics was the paradigm of inductive generalization.

Unfortunately, Hume’s critical examinations of logical justifications for the acceptance of inductive proofs were largely ignored [Russell, 1945 pp. 659ff.]. Consequently, most thinkers continued to believe that there was an inductive logic. Thus there was no apparent reason to doubt the claims made for the ‘scientific’ basis of Newton’s physics. And there was no reason to doubt the possibility of rational (i.e., inductive) decision-making. Supposedly, whenever one had all the facts, one needed only to be inductively rational to arrive without doubt at correct decisions. Moreover, whenever one made an error in judgment, it would have had to be due to either an irrational moment or a failure to gather all the facts.

Although economic theory has been deeply affected by the eighteenth-century beliefs about rational decision-making, the rationalism of economic theory is not obviously inductivist – with the possible exception of the textbook distinction between ‘positive’ and ‘normative’ economics. At least, very little of the faith in rationalism appears to have survived as explicit Inductivism. The reason for the absence of explicit Inductivism in mainstream economics today is that neoclassical economics reflects the concerns of late nineteenth-century and early twentieth-century philosophers, who were becoming aware of the possibility that Newton’s physics might not actually be true and, more important, that Inductivism might not be able to live up to its promises.

It can be argued that anyone who believed that Newton’s physical laws were true because they had been inductively proven must have been in some way mistaken. Such an argument would lead to two questions: (1) Did Newton fail to prove his theory true because he was mistaken about the objective quality of his ‘facts’? (2) Was Hume correct about the absence of an adequate inductive logic, so no quantity of ‘facts’ could ever prove Newton’s theory true? In response to such questions modern economic methodology falls generally into one of two opposing methodological camps depending on the answers given (what methodology is actually practiced by economists is a wholly separate question to be discussed later in this chapter). On the one hand (for want of a better name), there are the ‘conservative’ methodologists who would give an affirmative answer to (1) and a negative one to (2) and would promote the importance of the distinction between ‘positive’ and ‘normative’ economics. On the other hand, there are the ‘liberal’ methodologists who would give a negative answer to (1) and an affirmative one to (2) and would find the views of Solow and Samuelson, quoted above, more to their liking.

The Problem with Induction

The major point to be stressed here is that both positions taken by methodologists are based on Justificationism as well as on some form of Inductivism. And thus, both methodological positions accept the Problem of Induction. They differ only in regard to how the Problem with Induction is recognized.

The ‘conservative’ methodologists in economics say that there is nothing fundamentally wrong with inductive arguments, with the one possible exception that we must be very careful in the collection of ‘facts’. For the ‘conservative’ methodologists, if there should be a problem with the application of induction in economics or other social sciences, then it is that there are not enough ‘hard facts’ [e.g., Leontief 1971]. Specifically, before beginning an inductive proof one must assure quality and thus be careful to eliminate subjective or ‘normative’ opinions about what are the ‘facts’. The ‘conservative’ methodologists thus stress that for economics to be scientific it must be based on ‘positive’ rather than ‘normative’ statements.

The ‘liberal’ methodologists in economics take a position which is less optimistic but more devious. Rather than simply admitting that some theories which were once thought to be true are actually false, the ‘liberals’ obfuscate the methodological questions by denying that (non-tautological) theories could ever be true. For example, they might argue that only a tautology can be true and only a self-contradiction can be false [Quine 1965].

Theories, according to the ‘liberal’ methodologists, are to be considered ‘better’ or ‘worse’, rather than true or false. The reason for this switch is that the ‘liberal’ methodologists still think that the Problem of Induction must be solved before one can discuss ‘truth’ but, to their credit, they recognize that there is a problem with inductive logic. Specifically, they realize that no finite quantity of true singular statements could ever prove that any given general statement is true. In short, they admit that there is no inductive logic, and that is the Problem with Induction.

The retreat to Conventionalism

This doctrine [Conventionalism] contends that a scientific theory is, like a descriptive language, a device for ordering and communicating information which works because the members of a community know the rules and obey them. Thus, for example, in a telephone book all names are arranged in order according to the rules of the alphabet. This is purely a matter of convention. Any other ordering system could work equally well if it were generally accepted. The concepts of science, according to this view, are, similarly, only conventions that scientists have created. They are used to order empirical data but they cannot be construed to satisfy the positivist insistence that concepts should be representations of the real world.

This view of science has some merits. It emphasizes that science is a human creation and a social phenomenon, and it focuses on the utility of scientific concepts rather than their brute descriptive realism. But its defects greatly exceed its virtues. Like the contention that empirical observations are ‘theory-laden’, it considers only the nature of concepts, and neglects the role of explanatory hypotheses in scientific investigation.

Scott Gordon [1991, p. 610]

Despite the generous nods given to the positive/normative distinction in many economics textbooks, this popular distinction is nothing but a relic left over from late nineteenth-century attempts to save Inductivism (see J.N. Keynes [1917]). Since almost all economic methodologists have by now accepted that there is a Problem with Induction, one has to wonder why economics textbooks continue to promote the positive/normative distinction. The reason appears to be quite simple: For methodologists in economics, the Problem of Induction is still not dead!

The most openly adopted methodological position, in effect, puts Inductivism on a ‘back-burner’ for the present and temporarily puts a different position, ‘Conventionalism’, in its place along with Justificationism. I will argue here that, despite the attendant smoke, noise and celebration, the methodological controversies of the early 1960s, were merely family squabbles. That is to say, virtually all economic methodologists bow to the Problem of Induction (possible recent exceptions are Latsis [1972], Wong [1973], Newman [1976], Coddington [1979], Caldwell [1991a], Hands [1996] and Hoover [2001]). Since this problem is insolvable without an inductive logic, most methodological arguments in economics today are about the appropriate way to circumvent the Problem of Induction.

Given Conventionalism, it would appear that economists as methodologists do not attempt to solve the Problem of Induction itself but instead try to solve a weaker form of the Problem of Induction. For the purpose of discussing methodology, the problem-shift is unfortunate because the modified form of the Problem of Induction, which will be called the ‘Problem of Conventions’, is a bit more complicated than the original problem. The aim of the original Problem of Induction was a straightforward, objective, inductive proof of the (absolute) truth of any true theory. Contrarily, as I shall show, the aim of the Problem of Conventions is a choice of the ‘best’ theory according to current conventional measures of acceptable ‘truth’. Without an inductive logic, the solution to the Problem of Conventions can get rather complicated (in exactly the same way welfare economics has difficulties with social choices [see Boland 1989, chap. 5]). To add to the complications, there are many different measures to choose from (e.g., simplicity, generality, testability, etc.), and the measure used may or may not involve ‘inductive’ evidence.

The Problem of Conventions

Let me now state the problem which still dominates economic methodology.

The Problem of Conventions is the problem of finding generally acceptable criteria upon which to base any contingent, deductive proof of any claim to empirical ‘knowledge’.

Note that although the Problem of Induction and Problem of Conventions differ regarding the nature of the proof required for justification, they are the same in regard to the requirement of Justificationism. The word ‘knowledge’ has been specifically enclosed in quotation marks because one of the consequences of the presumed Justificationism is that ‘knowledge’ is not Knowledge unless it has been proven absolutely true, and deductive proofs always depend on given assumptions.

Where pure Inductivism requires a final (absolute) inductive proof for any true theory, Conventionalism requires o...