A particularly outstanding feature of J.B. Say's treatise is that he was the first economist to think deeply about the proper methodology of his discipline, and to base his work, as far as he could, upon that methodology. From previous economists and from his own study, he arrived at the unique method of economic theory, what Ludwig von Mises was, over a century later, to call ‘praxeology’. Economics, Say realized, was not based on a mass of inchoate particular statistical facts. It was based, instead, on very general facts (fait généraux), facts so general and universal and so deeply rooted in the nature of man and his world that everyone, upon learning or reading of them, would give his assent. These facts were based, then, on the nature of things (la nature des choses), and on the deductive implications of these facts so broadly rooted in human nature and in natural law. Since these broad facts were true, their logical implications must be true as well.
In his introduction to the Treatise, which sets forth the methodological nature and implications of his work, Say begins by being critical of the physiocrats and of Dugald Stewart for confounding the sciences of politics and of political economy. Say saw that if economics, or political economy, was to progress, it must stand on its own feet as a discipline without being intimately mixed from the start with political science – or the science which sets forth the correct principles of the political order. Political economy, wrote Say, is the science of wealth, its production, distribution and consumption.
Say goes on to mention the popularity of the Baconian method of induction from a mass of facts in the formation of a science, but then adds that there are two kinds of facts, ‘objects that exist’ and ‘events that take place’. Clearly, objects that exist are primary, since events that take place are only movements or interactions of existing objects. Both classes of facts, noted Say, constitute the ‘nature of things’, and ‘a careful observation of the nature of things is the sole foundation of all truth’.
Facts may also be grouped into two kinds: general or constant, and particular or variable. About the same time as Stewart, but far more comprehensively, Say then launched into a brilliant critique of the statistical method, and of the difference between it and political economy. Political economy deals with general facts or laws:
Political economy, from facts always carefully observed, makes known to us the nature of wealth; from the knowledge of its nature deduces the means of its creation, unfolds the order of its distribution, and the phenomena attending its destruction. It is, in other words, an exposition of the general facts observed in relation to this subject. With respect to wealth, it is a knowledge of effects and of their causes. It shows what facts are constantly conjoined with; so that one is always the sequence of the other.
Say then added an important point, that economics ‘does not resort for any further explanation to hypothesis’. In short, unlike the physical sciences, the assumptions of economics are not tentative hypotheses which, or the deductions from which, must be tested by fact; on the contrary, each step of the logical chain rests on definitely true, not ‘hypothetical’, general facts. (It might be added that it is precisely this crucial difference between the method of economics and of physical sciences that has brought so much contumely on the head of praxeology during the twentieth century.) Instead of framing hypotheses, economic science must perceive connections and regularities ‘from the nature of particular events’, and ‘must conduct us from one line to another, so that every intelligent understanding may clearly comprehend in what manner the chain is united’. ‘It is this’, Say concludes, ‘which constitutes the excellence of the modern method of philosophizing’.
In contrast, statistics exhibit particular facts, ‘of a particular country, at a designated period’. They are ‘a description in detail’. Statistics, Say added, ‘may gratify curiosity’, but they can ‘never be productive of advantage’ if they do not indicate the ‘origin and consequences’ of the collected facts and this can only be accomplished by the separate discipline of political economy. It is precisely the confounding of these two disciplines that made Smith's Wealth of Nations, in Say's perceptive words, an ‘immethodical’ and ‘irregular mass of curious and original speculations, and of known demonstrated truths’.
A crucial difference between statistics and political economy, Say goes on, is that the latter's general principles or ‘general facts’ may be discovered, and therefore may be known with certainty. The principles of political economy, wherever they rest on ‘the rigorous deductions of undeniable general facts’, ‘rest upon an immovable foundation’. They are what von Mises would later call ‘apodictic’. Political economy, indeed, ‘is composed of a few fundamental principles, and of a great number of corollaries or conclusions, drawn from these principles’. The particular facts of statistics, on the other hand, are necessarily uncertain, incomplete, inaccurate and imperfect. And even when true, Say correctly notes, they ‘are only true for an instant’. Again, on statistics, ‘how small a number of particular facts are completely examined, and how few among them are observed under all their aspects? And in supposing them well examined, well observed, and well described, how many of them either prove nothing, or directly the reverse of what is intended to be established by them[?]’ And yet the gullible public is often dazzled by ‘a display of figures and calculations... as if numerical calculations alone could prove anything, and as if any rule could be laid down, from which an inference could be drawn without the aid of sound reasoning’.
Say goes on to a blistering critique of the use of statistics without theory:
Hence, there is not an absurd theory, or an extravagant opinion that has not been supported by an appeal to facts; and it is by facts also that public authorities have been so often misled. But a knowledge of facts, without a knowledge of their mutual relations, without being able to show why the one is a cause and the other a consequence, is really no better than the crude information of an office-clerk...
Say then denounces the idea that a good theory is not ‘practical’, and that the ‘practical’ is somehow superior to the theoretical:
Nothing can be more idle than the opposition of theory to practice1. What is theory, if it be not a knowledge of the laws which connect effects with their causes, facts with facts? And who can be better acquainted with facts than the theorist who surveys them under all their aspects, and comprehends their relation to each other? And what is practice without theory, but the employment of means without knowing how or why they act?
Say then brilliantly points out why it is impossible for peoples or nations to ‘learn from experience’ and to adopt or discard theories correctly on that basis. Since the early modern era, he notes, wealth and prosperity have increased in western Europe, while at the same time nation-states have compounded restrictions of trade and multiplied the interference of taxation. Most people then superficially conclude that the latter caused the former, that trade and production increased as a result of the interference of government. On the other hand, Say and the political economists argue the reverse, that ‘the prosperity of the same countries would have been much greater, had they been governed by a more liberal and enlightened policy’. How can facts or experience decide between these two clashing interpretations? The answer is that they cannot; that only correct theory, theory deducible from a few universal general facts or principles, can do so. And that is why, notes Say, ‘nations seldom derive any benefit from the lessons of experience’. To do so, ‘the community at large must be enabled to seize the connexion between causes and their consequences; which at once supposes a very high degree of intelligence and a rare capacity for reflection’. Thus, to arrive at the truth, only the complete knowledge of a few essential general facts is important; ‘every other knowledge of facts, like the erudition of an almanac, is a mere compilation, from which nothing results’.
Furthermore, in arguments about public policies, when ‘facts’ are allegedly set against the ‘system’ of economic theory, it is actually one theoretical ‘system’ poised against another, and, again, only theoretical refutation can prevail. Thus, said Say, if you talk about how free trade between nations is advantageous to all the participants, this is accused of being a ‘system’, to which is opposed worry about deficits in the balance of trade – itself a system, but a fallacious one. Those who assert (as had the physiocrats) that luxury fuels trade whereas thrift is ruinous, are setting forth a ‘system’, and then, in an exact prefiguring of the Keynesian multiplier, ‘some will assert that circulation enriches a state, and that a sum of money, by passing through twenty different hands, is equivalent to twenty times its own value’ – also a system.
In a surprising and perceptive prefigurement of modern controversies, Say goes on to explain why the logical deductions of economic theory should be verbal rather than mathematical. The intangible values of individuals, with which political economy is concerned, are subject to continuing and unpredictable change: ‘subject to the influence of the faculties, the wants and the desires of mankind, they are not susceptible of any rigorous appreciation, and cannot, therefore, furnish any data for absolute calculations’. The phenomena of the moral world, noted Say, are not ‘subject to strict arithmetical computation’.
Thus we may know absolutely that, in any given year, the price of wine will depend on the interaction of its supply, or stock to be sold, with the demand. But to calculate the two mathematically, these two elements would have to be decomposed precisely into the separate influence of each of their elements, and this would be so complex as to be impossible. Thus:
it is not only necessary to determine what will be the product of the succeeding vintage, while yet exposed to the vicissitudes of the weather, but the quality it will possess, the quantity remaining on hand of the preceding vintage, the amount of capital that will be at the disposal of the dealers, and require them, more or less expeditiously, to get back their advances. We must also ascertain the opinion that may be entertained as to the possibility of exporting the article, which will altogether depend upon our impressions as to the stability of the laws and government, that vary from day to day, and respecting which no two individuals exactly agree. All these data, and probably many others besides, must be accurately appreciated, solely to determine the quantity to be put in circulation; itself but one of the elements of price. To determine the quantity to be demanded, the price at which the commodity can be sold must already be known, as the demand for it will increase in proportion to its cheapness; we must also know the former stock on hand, and the tastes and means of the consumers, as various as their persons. Their ability to purchase will vary according to the more or less prosperous condition of industry in general, and of their own in particular; their wants will vary also in the ratio of the additional means at their command of substituting one liquor for another, such as beer, cider, etc. I suppress an infinite number of less important considerations, more or less affecting the solution of the problem...
In short, the enormous number of imprecise, changing and quantitatively unknown determinants make the application of the mathematical method in economics impossible. And therefore those who
have pretended to do it, have not been able to enunciate these questions into analytical language, without divesting them of their natural complication, by means of simplifications, and arbitrary suppressions, of which the consequences, not properly estimated, always essentially change the condition of the problem and pervert all its results; so that no other inference can be deduced from such calculations than from formula arbitrarily assumed.
Mathematics, seemingly so precise, inevitably ends in reducing economics from the complete knowledge of general principles to arbitrary formulas which alter and distort the principles and hence corrupt the conclusions.
But how then is the political economist, knowing the general principles with certainty, to apply these principles to specific problems such as the condition of the wine market? Here, too, Say anticipated the brilliant conclusions of Ludwig von Mises on the proper relationship between theory and history, theory and specific application. Such applied theory in economics, Say indicated, is an art rather than a strict science:
What course is then to be pursued by a judicious inquirer in the elucidation of a subject so much involved? The same which would be pursued by him, under circumstances equally difficult, which decide the greater part of the actions of his life. He will examine the immediate elements of the proposed problem, and after having ascertained them with certainty (which in political economy can be effected), will approximately value their mutual influences with the intuitive quickness of an enlightened understanding, itself only an instrument by means of which the mean result of a crowd of probabilities can be estimated, but never calculated with exactness.11
J.B. Say then relates the fallacies of the mathematical method in economics to the teachings of his great mentor, the physiologist Cabanis. He quotes Cabanis on how writers on mechanics grievously distort matters when they deal with the problems of biology and medicine. Citing Cabanis:
The terms they employed were correct, the process of reasoning strictly logical, and, nevertheless, all the results were erroneous... it is by the application of this method of investigation to subjects to which it is altogether inapplicable, that systems the most whimsical, fallacious, and contradictory, have been maintained.
Say then adds that whatever has thus been pointed out about the fallacies of the mechanistic method in biology is a fortiori applicable to the moral sciences, which is why we are ‘always being misled in political economy, whenever we have subjected its phenomena to mathematical calculation. In such case it becomes the most dangerous of all abstractions’.
Finally, Say perceptively points to another problem that, then as now, leads learned people to dismiss the principles and conclusions of economics. For they
are too apt to suppose that absolute truth is confined to the mathematical and to the results of careful observation and experiment in the physical sciences; imagining that the moral and political sciences contain no invariable facts or indisputable truths, and therefore cannot be considered as genuine sciences, but merely hypothetical systems, more or less ingenious, but purely arbitrary.
To bolster this view, the critics of economics point to a great many differences of opinion in that discipline. But so what? Say asks. After all, the physical sciences have always been rent by controversy, sometimes clashing ‘with as much violence and asperity as in political economy’.
The mathematical method was not the only system of abstraction to suffer a trenchant demolition by J.B. Say. For Say was also sharply critical of verbal methods of logic that took off into the empyrean without continuing groundwork in, and repeated checking by, reference to general and universal facts. This was Say's main methodological stricture against the physiocrats. ‘Instead of first observing the nature of things, or the manner in which they take place, of classifying these observations, and deducing from them general propositions’ – that is, instead of being praxeologists, the physiocrats
commenced by laying down certain abstract general propositions, which they styled axioms, from supposing them to contain inherent evidence of their own truth. They then endeavoured to accommodate the particular facts to them and to infer from them their laws; thus involving themselves in the defence of maxims evidently at variance with common sense and universal experience...
In short, a system of economic theory must not only be axiomatic-deductive; it must always make sure to ground those axioms in ‘common sense and universal experience’.
In his Introduction to the fourth edition, Say levelled similar strictures against David Ricardo and the Ricardian system. Ricardo, too, ‘sometimes reasons upon abstract principles to which he gives too great a generalization’. Ricardo, he charged, begins with observations founded on facts, but then ‘pushes his reasonings to their remotest consequences, without comparing their results with those of actual experience’. After a certain point in the reasoning, ‘the facts differ very far from our calculation’ and ‘from that instant nothing in the author's work is represented as it really occurs in nature’. ‘It is not sufficient’, Say concludes, ‘to set out from facts; they must be brought together, steadily pursued, the consequences drawn from them constantly compared with the effects observed’, so that
the science of political economy... must show, in what manner that which in reality does take place, is the consequence of other facts equally certain. It must discover the chain which binds them together, and always, from observation, establish the existence of the two links at their point of connexion.