Hostname: page-component-78c5997874-s2hrs Total loading time: 0 Render date: 2024-11-04T15:08:26.688Z Has data issue: false hasContentIssue false
  • English
  • Français

Notes on the Verifiability of Economic Laws

Published online by Cambridge University Press:  14 March 2022

Emile Grunberg
Emile Grunberg*
Affiliation:
University of Akron
Get access Rights & Permissions [Opens in a new window]

Extract

To-day economics is considered the most highly developed discipline among the social sciences. Yet, its explanatory and predictive powers are admitted by all hands to be weak compared to those possessed by the physical sciences. This weakness is still frequently explained by apologetic references to the relatively tender age of economic science. This apparently implies that in the normal course of time economics will grow up and achieve the stature and powers of say, physics. The trouble with this argument is that economics is certainly as old as physics. If it lags behind other disciplines in predictive and explanatory powers, lack of time cannot be the cause.

Type
Research Article
Information
Philosophy of Science , Volume 24 , Issue 4 , October 1957 , pp. 337 - 348
Copyright
Copyright © 1957, The Williams & Wilkins Company

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Footnotes

†

I am especially endebted to Professor Carl G. Hempel who pointed out to me many errors in my argument. My thanks are also due to Professors Martin Bronfenbrenner Franco Modigliani, and Herbert Simon, as well as to several other colleagues too numerous to mention here. Their criticism and advice were of great help to me, but the responsibility for the argument as presented here is entirely mine.

References

1 In modern times, for example, Nicolas Copernicus contributed an important essay to monetary theory (four versions between 1519 and 1529) of which the last version, Monetae Cudendae Ratio, is best known.

2 Professor C. G. Hempel suggested to me two definitions of the “unique event”. First, the event might be understood as all that goes on in a specified space-time region. Second, the event might be considered as the occurrence of a specific attribute in a specified space-time region. In this paper the term “unique event” always refers to the first definition. The term “historical event” is used synonymously.

3 To the extent that these disciplines deal with unique events, it is in the sense of Hempel's second definition (cf. note 2).

4 Cf. M. Brodbeck, “On the Philosophy of Social Science” (Philosophy of Science, 1954), pp. 146–47. Also: E. Grunberg, “Notes on Historical Events and General Laws” (Canadian Journal of Economics and Political Science, 1953), pp. 512–14.

5 Every observable admitted via a specific observation into O and O∗ must appear in at least one law in L.

6 What is regarded as an event is arbitrary. At one time it may be the price trend on the stock-exchange over several weeks or months; at another time it may be the precise number of transactions in a particular stock on a particular morning.

7 “The Function of General Laws in History” (Journal of Philosophy, 1943, reprinted in H. Feigl and W. Sellars, Readings in Philosophical Analysis, New York 1949). The explanation-sketch seems to correspond to what in the language of econometrics is called a structure known to be incomplete. In econometric procedure the incompleteness is remedied by the inclusion of stochastic variables representing the combined effect of variables excluded from the explanation. Cf. J. Marschak, “Measurement for Policy and Prediction” (W. C. Hood and T. C. Koopmans, Studies in Econometric Method, Cowles Commission Monograph No. 14, New York, 1953) p. 12.

8 Generally this is not a single decision but a sequential process continued until criteria stipulated by the researcher are met. The choice of the criteria raises fundamental problems which, however, transcend the scope of this paper.

9 Cf. M. Friedman, Essays in Positive Economics (Chicago 1953), pp. 8–9.

10 Sometimes the explanation of a unique event is said to make possible the prediction of a subsequent event. Let these events be E and E' respectively. Closer examination reveals that—regardless of whether there is an explanation of E—the prediction of E' simply uses observable attributes of E as initial conditions. Likewise the set L' used in the prediction of E' is independent from the set L which might be used in the explanation of E.

11 E. Zilsel in “Physics and the Problem of Historico-Sociological Laws” speaks of historical laws. What he means are generalizations derived from the observation of historical occurrences. Like all general laws, these too assert invariant relationships between variables and thus refer to classes of events. (Philosophy of Science, 1941, reprinted in H. Feigl and M. Brodbeck, Readings in the Philosophy of Science, New York 1953, especially pp. 718–22).

12 J. B. Conant (On Understanding Science, Yale 1947, p. 36) suggests that theories are overthrown by better theories and not simply by contradictory observations. Cf. also F. Machlup, “The Problem of Verification in Economics” (Southern Economic Journal, 1955), p. 11. It is interesting that Conant uses the term “overthrown” instead of “disconfirmed”. His statement does not seem to deal at all with the rules of methodology, but rather with the observable behavior of scientists. It is a psychological observation and as such in the object-language, and not in the meta-language. Cf. also N. Campbel, What is Science? (New York 1952), pp. 64 ff.

13 Usually events under controlled laboratory conditions are considered to occur in closed systems.

14 Verification need not be prior to the use of a law in an explanation-sketch. It is permissible to use untested hypotheses. The point is that such hypotheses must be subjected to verification independently from the specific application. If the verification of a general hypothesis is undertaken after its use in an explanation-sketch, this sketch will be considered tentative, or—better—the degree of our confidence in it will be low.

15 Cf. J. S. Mill, Essays on Some Unsettled Questions in Political Economy (London School of Economics and Political Science Reprint No. 7), pp. 147–50. Especially in German social thought this fact served as basis for the assertion of a fundamental difference be- tween the phenomenal domain of the natural sciences and that of the mental or moral sciences (Geisteswissenschaften). Among the philosophers F. Rickert must be remembered here. Among the social scientists such names as M. Weber and W. Sombart come to mind. On the other hand, neoclassical theory asserts that no major conceptual difficulties are created by the economist's inability to perform isolating experiments. (Cf. M. Friedman, op. cit., pp. 10 ff; F. Kaufmann, Methodology of the Social Sciences, New York 1944, p. 177: V. Pareto, Manuel d'Economie Politique, Paris 1927, pp. 15–17). The often used reference to astronomy and meteorology, however, offers little comfort: astronomy is conceived as a closed system and meteorology is beset by the same difficulties as economics.

16 The difficulties hidden behind the word “effectively” transcend the scope of this paper.

17 Cf. F. Machlup, op. cit., pp. 17–18; also, for example, A. H. Hansen, “More on the Multiplier Effect of a Balanced Budget” (American Economic Review, XLVI, 1956) pp. 157–58.

18 The following passage may serve as a somewhat exaggerated illustration of this point:

“All these forecasts, it should be noted, are based on two important assumptions—that there will be no major depression and no new wave of inflation. If either of those unhappy events should occur, all bets are off.”

(New York Times, January 8, 1956; the passage refers to a forecast of capital outlays by the Machinery and Allied Products Institute). An economic forecast made under the condition that neither a depression nor an inflation occur, is of little predictive power.

19 This reasoning is of the form of denying the consequent. It is formally valid.

20 The term “law of demand” is used here to designate the general form of the demand function which is usually written:

where Di is the quantity demanded of the i's commodity, the pi's stand for the prices of all goods and services, and R stands for income.

21 Cf. Sidney Schoeffler, The Failures of Economics: A Diagnostic Study (Harvard University Press, Cambridge 1955), pp. 76–82 (The Micro Theory of Demand) and, in general, the penetrating analysis of major pieces of economic theory given in the case-studies of chapters 5 and 6.

22 Cf. T. W. Hutchison, “Professor Machlup on Verification in Economics” (Southern Economic Journal, XXII, 1956), pp. 481–82.

23 Cf. H. Bernardelli, Die Grundlagen der Oekonomischen Theorie (Tübingen 1933), pp. 6–7. See also: “On the Economic Principle, A Correspondence between B. Croce and V. Pareto, 1900–01” (International Economic Papers, No. 3, New York, 1953), especially Pareto: “We have at last come to see that a certain science called pure economics, … studies only a limited fraction of the economic phenomenon as you have defined it, and studies it in a particular manner, i.e. exclusively by deduction from one or from a few hypotheses, from one or very few facts.” (p. 185). L.v. Mises' Praxeology encompasses a great deal more than Pareto's pure economics. Its theorems are obtained by deduction from premises claimed to be somehow a priori true and thus exempt from the need for verification. Thus, according to Mises: “Praxeology conveys exact and precise knowledge of real things.” (Human Action, Yale 1949), p. 39; see also: pp. 51 and 64–71.

24 Op. cit., pp. 14–15 and 16–23.

25 Op. cit., pp. 10–11. Cf. also: J. R. Hicks, A Revision of Demand Theory (Oxford 1956), p. 17.

26 If I correctly understand Professor Friedman's position, he thinks that even though in most social sciences controlled experiments cannot be performed, experience “casts up” evidence which is “abundant and frequently as conclusive as that from contrived experiments …” (Op. cit., p. 10). This evidence is said to make possible the verification of theorems deduced from the rational principle. As has become clear by now, I disagree on this point with Friedman's otherwise excellent clarification of some of the main methodological issues of economics. He is correct in arguing that no experiment is ever completely controlled and that experience in some vague sense is not entirely uncontrolled. However, the laboratory experiment contrives precisely to isolate those factors which in the light of available knowledge and of the hypothesis to be tested are most likely to affect the outcome. As to Friedman's reference to our historical experience with inflation (p. 10), I believe it to be not very relevant to the point at issue. The assertion of a relationship between the volume of money in circulation and the price-level is a crude generalization (known already to Jean Bodin in the sixteenth and to some extent also to Nicolas Oresme in the fourteenth century). It seems compatible with and deducible from a variety of behavior assumptions. It is also subject to the ceteris paribus condition.

27 “In order to be of service …, however, economic theory will have to build upon ‘lawful’ postulates and not upon convenient but counterfactual assumptions.” (S. Schoeffler, op. cit., pp. 84–85). Cf. also the now famous “marginalist controversy” of a few years ago in which participated Professors R. Lester, F. Machlup, H. M. Oliver, R. A. Gordon (all in the American Economic Review, between March 1946 and June 1948, vols. XXXVI to XXXVIII).

28 Cf. F. Machlup, op. cit., p. 19; also J. Marschak, op. cit., p. 14.

29 Cf. F. Machlup, op. cit., pp. 21–22; see also S. Schoeffler, op. cit., pp. 101–2. Professor Schoeffler's use of the term “observation” to express the same idea (e.g. pp. 64 and 72) leads him into confusing the explanation of a lower-level law by a higher-level law with the lower-level law's verification. If I correctly understand him, he argues that intended empirical laws in economics (only?) “cannot claim to specify how things must necessarily happen in the future” (p. 72), because they do not indicate the causes of what has been observed to happen. Now, lower-level laws are explained by higher-level laws from which they are validly deduced—but this neither verifies them nor increases their own predictive power. On the contrary, higher-level laws are indirectly verified by the verification of the lowest-level (empirical) laws which can be deduced from them. Verification, on the other hand, we are agreed, requires prediction of observations. (Cf. Section 5, footnote 33). The same confusion seems to underly Sehoeffler's argument:

“The malperformances of the theory in actual practice provide additional evidence of its undependability, but this further evidence is not necessary. The methodological evaluation is sufficient.” (p. 95).

As Schoeffler himself argues: the only test of a theory is its performance, which is explained and predicted by methodological evaluation.

30 Cf. above, pp. 341 ff.

31 The difficulties connected with the concept of the degree of confidence with which a hypothesis is held, are central to the methodology of science. A discussion of these issues transcends, however, the modest scope of the present paper.

32 It need no longer be argued that neither the collection of data, nor even the loosest narration of events are conceivable without a theoretical framework. Cf. Louise Sommer, “Zum Wirklichkeitsgehalt ökonomischer Theorien” (Schweizerische Zeitschrift für Volkswirtschaft und Statistik, vol. 82, 1947), pp. 124–30.

33 The fact that a law is deductible as theorem from a higher-level law may enhance our confidence in it; but its explanatory and predictive power depends solely on the correspondence of the law's assertion with actual observations. Cf. E. Zilsel, op. cit., p. 175.

34 See above, Section 3.

35 Cf. F. Kaufmann, op. cit., pp. 39, 53, 112, 231.

36 “Prediction and Hindsight as Confirmatory Evidence” (Philosophy of Science, July 1955).

37 As object of scientific research expectations are, of course, also overt actions. Cf., for example, An Appraisal of Data on Businessmen's Expectations about Outlook and Operating Variables, Report of Consultant Committee on General Business Expectations (Board of Governors of the Federal Reserve System, Washington, D. C., 1955) and the literature listed there.

38 Cf. Gustav Bergmann, “Reduction” (Current Trends in Psychology and the Behavioral Sciences, Pittsburgh 1954); and M. Brodbeck, “On the Philosophy of Social Science” (Philosophy of Science, 1954).

39 Cf., for example, J. S. Mill, op. cit., pp. 133–41. C. Menger calls economics “a difficult and previously unexplored field of psychology” (Principles of Economics, Glencoe, Ill., 1950). P. H. Wicksteed in his contribution to Palgrave's Dictionary of Political Economy, “Political Economy and Psychology” (vol. III, London 1918, pp. 140–42) treats economics as a field of applied psychology. In recent years, for example, Frank H. Knight and F. Hayek, have emphasized the fundamentally psychological nature of the phenomenal domain of economics (cf. F. Hayek, The Counter-Revolution of Science, Glencoe, Ill., 1952). On the other hand, from outside the current of neo-classical thought, W. Sombart severely criticises “psychologism” in economics which he views as a persistent and serious error (Die Drei Nationalökonomien, München und Leipzig 1930, pp. 161–67). From within J. N. Keynes asserts that economics presupposes psychology, yet is itself not a part of this discipline (The Scope and Method of Political Economy, London 1891, p. 85). Similar ideas are also expressed by F. v. Wieser (Theorie der Gesellschaftlichen Wirtschaft, second edition, Tübingen 1924, pp. 8–12). It is not quite clear whether this latter attitude is different from that of, say Menger and Wicksteed, and ultimately from the one expressed here, in essence or merely in form. To assert that discipline A is presupposed by discipline B implies that propositions formulated in B are reducible to those formulated in A.

40 Cf., for example, C. A. Hickman and M. H. Kuhn, Individuals, Groups, and Economic Behavior (New York 1956). Work like that of Professors R. Cyert and J. March ought also to be mentioned here (cf. “Organizational Structure and Pricing Behavior in an Oligopolistic Market”, American Economic Review, XLV, 1955).

41 Cf., for example, E. M. Thrall, C. H. Coombs, and R. L. Davis (editors), Decision Processes (New York 1954).

42 Not all laws used in economic explanation-sketches are economic laws. The economist may use laws from every discipline. Nor are all of these borrowed laws strictly disconfirmable. The economist, however, is responsible only for his own laws, and for informing himself about the status and the reputation of the laws which he borrows from other disciplines.

43 The argument in the text neglects such contributing difficulties as are, for example, encountered currently in collecting data.