press enter after type

USDA Graduate School. Frank Knight Lecture on Economics Methodology, 1930


In an obscure publication of a series of special lectures at the United States Department of Agriculture held in 1930, I found the following interesting methodological reflections of Frank Knight that are reproduced below. An earlier post provided E.B. Wilson’s thoughts on the application of scientific methods in economics (see link below) which more or less staked out precisely the opposite position to Knight. 




Contents: The following lectures were delivered before the students of the Graduate School in February and March 1930, and are issued in this form for present and former students of the school.

Scientific Method in Economic Research
by Dr. E. B. Wilson, President, Social Science Research Council.

Evaluating Institutions as a Factor in Economic Change
by Prof. John R. Commons, University of Wisconsin.

Analytical Methods in Agricultural Economics Research
by Dr. John D. Black, Harvard University.

Fact and Interpretation in Economics
by Dr. F. H. Knight, University of Chicago.



By Dr. F. H. Knight, University of Chicago.

My task on this occasion is one to be approached with misgivings, and I do approach it with doubts. I do not see clearly and surely in the field of economic methodology, and the airing of doubts, or viewing with alarm is likely to be thought an ungracious performance. Nobody loves a bear! But after all doubts have their place. We do not get where we want to be by driving with enthusiasm and power and speed in the wrong direction. And I do feel strongly that some present trends in economic activity carry more than a threat of wasted energy. If the effort to solve a problem is to be fruitful it must be put forth in the light of a correct conception of the nature of the problem itself, and there can be no real gain from conceiving a problem more simply than it realty is, and thus make the solution appear easier.

My reference is of course to the current enthusiasm for making the study of economics “scientific,” meaning factual, concrete and quantitative, or specifically, statistical. I have to raise questions and suggest doubts as to whether the proper content of this study, or “science” can really be facts, whether it can really be a “science” if we use the term in the sense it carries in speaking of the natural sciences. As the subject announced is intended to suggest, I must argue that Economics deals rather, primarily, with meanings with what facts mean rather than facts themselves. Consequently, while of course we have to consider facts and be careful to get them “right” we have to approach them, and look at their rightness and wrongness in very different terms from those proper to the natural scientist; for the economist or other social scientist, in this view, facts are preliminary, not the real subject matter of the study. The main theme of these remarks will then be the contrast in character and method between the natural sciences and those which deal with man in society, with particular reference, of course, to economics.

At the outset, however, I want further to say that I understand the feelings of those who want to make economics an objective and quantitative science, and sympathize with them deeply. The “backwardness” of the studies dealing with man, in comparison with those dealing with nature, is superficially an obtrusive fact, and one which seems superficially to point its own moral. In the face of the contrast between the solid achievements of the natural sciences in the past few centuries, and the relative lack of advance in the understanding or control of social relations since the Ancient Greeks, it is natural to conclude that the way to reform the social sciences would be to imitate those which appear so much more successful in their task. And in particular, it is natural to hit upon the theory that the social sciences have “remained” in the “speculative” stage, while the natural sciences have taken to careful detailed observation, measurement and experimentation. In the face of this situation, to repeat the thought in more vernacular terms, it is most natural to develop a certain impatience, to insist on getting out of the stage of speculating and arguing what to do, and do something, and to put content into this by making it mean to get the facts, bring them into relation with each other and see how they may be used for prediction and control, as the physical sciences have been so successful in doing.

However, a little examination will show that the case is not so simple as that. To begin with, we have long had natural sciences of man and they tell us nothing about social events. The physics, chemistry, biology, physiology and pathology of the human organism are extensively studied and well developed and beyond a few broad and obvious statements, mostly negative, they do not reveal anything about the course of history, or make possible the prediction and control of social movements. We know that human beings will always eat, and that if they live in certain climatic zones they will have some protection from the elements. Perhaps we may add speech and recreation as biological traits. But such general information is of no concrete use to the economist, for example. To be useful to him it must go so much farther, into so much greater detail, as to what people will eat, wear, etc., and how much, and how, that the problems become different in kind as well as degree. As soon as we try to make general statements in this field, we find that any general import they have runs in terms of something quite other than the facts observed by the senses. The uniformity, as suggested already, is in the meanings, not in the concrete content of behavior. Even in the matter of food, it is men’s knowledge or beliefs about what is desirable or “fit” to eat rather than that actual physical qualities of materials which are decisive.

The best illustration in principle is in the field of communication. The sounds and characters are physical facts, but there is practically no discoverable relation between these and what they are used to convey. If we know anything for sure, we can say we know there is no connection between language differences and either physical differences in the peoples or the content of thought or emotion they wish to communicate. It appears that any person could equally well learn any language and, that with slight reservations, not important in this connection, any language can equally well express any content that is expressible.

The function of the natural sciences is to describe the properties and “behavior” of things as they appear to our senses, that is, physical things and materials in space, and behavior which reduces to rearrangement of matter in space. The essence of it is the descriptive point of view. It tells what happens, not why anything happens. From the “pure” science point of view itself (separated from practical significance) it enables us to understand the complex manifold of events in the outer world by reducing them to a manageable number of elemental general principles, especially and perhaps at last entirely, those of mechanics. It does this by finding “uniformities” or “repetitions” in events, by showing that under similar conditions similar consequences follow. Thus Newton showed that the movements of the heavenly bodies exemplify the same phenomenon of “falling” that is familiar for objects near the earth’s surface; and Darwin showed that the production of the infinite variety of plant and animal forms might be viewed as a working out of the same principle as the production of new varieties through selective breeding by the gardener or fancier.

Back of this function of science of enabling us to understand things, of explaining and so satisfying our intellectual cravings, is, as we all know, the practical function or functions, of making possible prediction and control. The fundamental point here, which seems to be overlooked in proposing to make the social sciences “scientific” is that the natural sciences themselves are based on the assumption of a sharp antithesis between man and nature. Man is the controller, nature the to-be-controlled. In fact, quite aside from this practical relationship between user and used, workman and tool, the same insuperable opposition really holds in the mere logical relationship between knower and known, or understander and thing, or matter understood. But it is clearest in the practical view. All our notions of prediction and control, by man over nature, through science are bound up in a conception of nature as passive, over against ourselves as possessed of mind, will and initiative. It is never trying to control man. More specifically, we view nature as an aggregate of things and materials in space, purposeless and inert in themselves, completely amenable to “control” from without in the particular sense of being movable from one place to another, which movement may liberate potential energy stored up in them, or modify the process of storing up or releasing such energy in some way.

When we examine the notion of prediction we find that it reduces either to the fact of “inertia,” the property of things by which they stay where they are or keep on moving as they are moving at any time, unless “acted upon” in the sense of having motion (or some new motion) imparted to them from without, or to the release of potential energy. The notion of control is always relative to movement because the only way in which human beings can act upon the external world or produce any change in it is through our voluntary muscles, which can directly produce only the, change of moving some bit of matter from one point in space to another. All changes which man produces and which constitute his “control” over nature are the results in nature of such movements of matter if they go beyond the immediate fact of motion itself. Most of our knowledge of nature, the content of the sciences, which gives variety and significance to our control activities, consists of facts regarding the processes (always the same under the same conditions) according to which energy is stored up in or released from natural materials in connection with their spatial relationships. The amount of energy communicated to natural objects by our muscles directly is generally negligible, though such a movement as striking a match may start energy changes which will explode a magazine or burn up a city.

The point here is merely that science itself depends on the assumption that just as things do not move or change their state of motion of themselves, they do not change their behavior in storing up or releasing energy of themselves, but do change as to these processes in uniform ways in response to outside acts of the form of moving them about in space in relation to each other. These uniformities are physical. A natural process, for instance, may be set off by a sound. It is said that avalanches have been started by sound waves. But in nature, the same sound will always produce the same effect. Sounds, and other causes, act as what they physically are, and not as symbols or bearers of meaning. Let us consider the contrast between this situation and that presented by the problem of applying scientific method in the field of the study of man.

In the first place, we must again note, human beings are undoubtedly natural objects, things in space, and as such they seem to be subject to all the laws and principles which science finds to hold for other objects under the same conditions. The same principles of physics and chemistry and physiology apply in the human body as elsewhere, as far as the most careful measurement reveals. But in addition some other principles seem to apply which do not hold good elsewhere. Men are more than mechanical objects which release energy in uniform ways in response to external movements of matter. They initiate changes, out of all discoverable uniformity of relation to external changes of any kind; and when they do respond to external changes, the nature of the response has relatively little uniform relation to the physical nature of the stimulus but is chiefly a matter of what we call the meaning of the stimulus-event which puts the whole occurrence, as the philosophers say, in a different universe of discourse. These meanings and the responses to them depend on the history, which is a thing made up of meanings, of social groups and the particular life-history of the individual in the group; and they are very largely free from “dependence” on anything which research has yet disclosed. As far as can be judged in the present state of knowledge (in the speaker’s opinion) the problem of understanding and explaining these phenomena must be approached in a quite different way from that of understanding and explaining physical nature. (In the scientific sense I mean; ultimately, philosophically, the problem of explaining nature is itself likely very different from that of science, for as already noted science does not pretend to give any answer to any question of why things are as they are.)

The root of the difficulty in regard to explaining and controlling human beings is the fact that the explainers and controllers are likewise human beings. It is impossible to regard human beings as of one kind when understanding and exercising control and of another and totally different kind when being understood and controlled and yet the two roles call for different characteristics. I shall return to this point presently. For the moment I wish to go a little more into detail about the “more,” in the statement that man is more than an object in space behaving in relation to other objects in accordance with universal mechanical principles.

It is possible to look at a human being in several strongly contrasting ways, and describe him in different sets of terms. We may look at him, for example, in psychological terms, and “explain” his acts by relating them to mental states. Many changes can be wrung on this theme. The philosopher Hegel gave a logical or dialectical interpretation of history, and the British psychologists of the early nineteenth century explained human nature in terms of association of ideas.

Another possible approach is in terms of “institutions,” a term which is being much used in economics these days, and very loosely used, and largely misused. An institution in the proper sense is a phenomenon of the nature of the language. It is neither a mechanical response to a physical stimulus nor a deliberately contrived procedure for achieving an end. Language is of course a tool, it is seen to be one after it has developed, but no one ever contrived it (in so far as it is a pure type of institution). It is believed by students of the subject that language actually developed primarily as a vehicle of emotional expression and acquired its more utilitarian functions secondarily. In any case, the methodological point is that the student of language treats it as an entity on its own account, indeed without very express reference to human beings or their interests and acts. It seems to have its own laws of relationship and of change, much like an organic species. It is a figure of speech, but a descriptive one, to call the human group the soil in which a language, or other institution, grows. Just as the plant one gets depends on the seed sown and not to any great extent on the soil, so it seems that institutions grow and change without much reference to the human beings who carry them on — though sensitive to contact with other related institutions with which they may hybridize, again much like plants.

There is much justification for an “institutional” approach to what we call economic phenomena. If we look at the facts of wealth and the processes of its production, distribution and consumption, and ask “why” these things are as they are, it is a very defensible answer to hold that they are customs which have grown up, much in the way in which a language grows, and to be “explained” only by giving the details of the history of that growth. Such an interpretation should, it seems clear to me, be kept very distinct from the “statistical” approach to the same problem. Economic statistics stand as a method at the opposite pole from institutional history. There is little or no distinctly human content of any kind in them. They relate almost entirely to commodities as such, and to external means of economic life rather than that life itself.

It is to be noted that the traditional or orthodox economic thought, in the British utilitarian line, is very different from both of these; in fact institutionalism and business statistics represent reactions in opposite directions from the utility-and-cost, supply-and-demand economics. The conception of human nature involved in the latter is interesting and needs to be clearly understood. Man is not looked on as a physical behavior mechanism, or a psychological being, or as the bearer of institutions, but as a being who has wants and limited means for satisfying them, and who is confronted with the problem of making the means go as far as possible. The means and ends of action are data, the procedure itself problematical. This standpoint will be clearer if it is contrasted, on the one hand, with a mechanical view of human nature, in which the response is completely determined by the conditions and hence is not in any sense problematic, and, on the other hand, with a view (or with a type of situation) in which action is conceived in terms of means and end but the end is also conceived of as problematical. As I myself see the matter the view of “unsophisticated common-sense” is in the main that of the classical economics. We assume that people in general know what they want, and are confronted with the problem of getting it, in the maximum degree, with the limited means at hand, which problem they “solve” more or less completely, through intelligence or luck. The problem itself, the ends to be realized and the means and conditions are given in the person and his situation, but his activity in “solving” it is peculiar in that it involves effort and in general a greater or smaller margin of error, these being absent from mechanical reactions.

When we look critically at human behavior, it seems to me that we are forced to recognize that the ends of action are problematic in about as great a degree as the means. Life seems to be an exploration as much as it is a quest in which we know what we are trying to find. This conception might be designated by speaking of the ethical man, in contrast with the economic man and the mechanical or behavioristic man, a variation of which would be the institutional man.

The difficulty is that all these views, and still others which I cannot here even list, have some degree of validity, and yet it is most difficult to make them seem consistent with each other. The philosopher Kant gave effective statement to a part of the problem, the conflict between the mechanical and ethical view of human nature, in his famous statement that man is at once subject to universal causality and a self-legislating member of a kingdom of ends. As I see the “facts” – which are facts in the sense that everyone treats them as such when he is not expressly trying to prove some theory – the situation is much more complicated, and hence much “worse” from the standpoint of our intellectual cravings and practical needs for simplicity. We seem to have to reconcile ourselves to the fact that man is at once not merely two but a great many different kinds of being, kinds which seem logically contradictory. He is different kinds under different circumstances, or capriciously or accidentally, and he is even several kinds in the same situation. He is a cause-and-effect mechanism and a bearer of culture or “soil” in which institutions grow according to their own laws of growth, a being of irrational judgments and a being who deliberates and decides intelligently (more or less!) and this both regarding procedures for reaching ends which he accepts unconsciously and also about ends to be chosen and pursued. For anything like completeness we should have to add still other items to the list, such as that he is commonly and in all sorts of degrees a dreamer and mystic and even an intrinsically “contrary” being and often takes a perverse delight in being thwarted and punished and in having grievances against the world and all and sundry in it.

It is indeed a formidable if not forbidding task to theorize about such a creature or formulate generalizations in terms of which his actions can be predicted and controlled. But it is hardly in conformity with the scientific attitude to insist on false simplification or refuse to face the facts because they present difficulties. The contrast between the problem of prediction and control in the case of a mechanism and in the case of human beings may be seen in a number of kinds of simple illustrative cases. In the first place, the entire theory of science depends, as noted above, on the repetitiveness of events and uniformity of relationships; the same effects follow the same causes. But in the mere external facts of the case this is not true of human beings. Physically, chemically and physiologically they are alike, enough to infer from one case to another, within limits, though it must be remarked that even in this field the science of medicine is seriously embarrassed by unaccountable differences in the reaction of different cases to the same treatment. Moreover, the doctor, if candid and shrewd, relies perhaps as much on psychological treatment wisely varied to fit the case as he does on drugs and physical therapeutic agents. On the plane of social behavior, however, even this minimum of uniformity seems conspicuously absent. Experiment with one human being simply does not tell how another will respond to the same experiment, as nearly identical as it is possible to make the repetition.

And worse, it is in the very nature of the creatures that the same one will not ordinarily respond in at all the same way if an experiment is repeated. Let anyone try the simplest experiment, such as telling another a story or sticking him with a pin or offering him a present of a five dollar bill, and then repeat the “stimulus.” It is, as just stated, the very nature of a human being not to be at all the same person with reference to a repeated situation as to its first occurrence. A gun or a trap which has been discharged or sprung is, when reloaded or reset, the same as before, but you cannot restore a person to the original condition, even to the degree within which it is possible to find another like him. People are different from mechanical objects in that they have a history. In part this difficulty may be avoided by taking them in groups, but groups also are always unlike and each group has a history. None of us is like his forefathers, even in the tenuous sense in which he is like his contemporaries. Our “situations” are very different, and our responses are different even where the situations appear similar.

This does not mean that the case is hopeless, that there is no place for intelligence in human relationships, or even that it is impossible to effect improvement through diligent observation and study. Our everyday experience proves the contrary. With all our bewilderment, we do have a fair knowledge of what to expect of our fellow-beings in ordinary situations and of how to treat them to secure cooperation and orderly living. It is a question of method. We do not acquire our common-sense knowledge of how to get along with our fellows in the same way as our common-sense knowledge of how to respond to and use natural objects, and it is reasonable to suppose that in the one case as in the other improvement will be secured by refinement along the general line of common-sense procedure. The essential fact in understanding our fellow human beings is primarily that we communicate with them. Thus in a sense we get inside of them instead of merely observing them from without. Of course our communication is based upon external observation, but the essential difference remains.

It is impossible to elaborate upon this difference here, and it should not be necessary. The heart of it is the contrast between a more direct instinctive but unformulated knowledge, based on familiarity on the one hand, and, on the other, reduction to rule in terms of physical units. A good illustration is the learning of a language. We can and do, without great difficulty, learn the meanings of sounds and characters and recognize them with fair accuracy and with little effort. But to base such knowledge on physically measured specifications as to the precise wave-forms or shapes would be quite out of the question practically, though a certain amount of such study may be interesting afterwards. The principle holds throughout the field of human phenomena and relationships. We describe people and works of art and literature and other products with a fair degree of intelligibility, and recognize them by their traits, though we could not make a beginning at putting this knowledge in accurate, scientific, physical terms. (Of course the artist who wishes to simulate effects in a physical medium does have to know in a sense how the lines and colors go, but his knowledge is also an immediate feel of how to do the thing and nearly as far remote from the ideal of mechanical “directions” as is the interpretative recognition of the layman.

My concrete suggestion is that if economics and the social sciences want to make more rapid progress they must give up the visionary ideal of building a society from blueprints and dimensions as we build a house and quit trying to imitate engineering and the sciences upon which it is based and turn rather to the study of their own data and. the processes by which we do come to have some intelligence in relation to these data on the level where progress, has already been achieved. That is, we should learn from “art” in the broad sense, and from the way in which the arts are learned and taught rather than from physical science and engineering technique.

It is to be admitted that in an important sense this is less satisfying. Our minds to crave the definite rule, the fool-proof formula. But it is a question of facing facts, and the actual character of the problem. It will never be as simple and definite a matter to improve the grammar or the morals of a social group as it is to build a bridge or compound a chemical. But we shall not make the task easier by insisting on applying methods which would admittedly be more satisfactory if they could be applied but which simply will not work because it is not that kind of a problem.

In conclusion I wish briefly to call especial attention to two sets of facts. The first is that in controlling human beings the “techniques” employed include such things as teaching, persuading, exhorting, or finally deception and coercion (which may presumably be practiced for “good” as well as “bad” ends). The point is that such concepts hove no meaning in connection with the procedure for controlling physical objects. When these procedures are sometimes applied to the higher animals it is evident that we are treating them like human beings rather than like mechanisms.

The second fact, or set of facts, is closely related to the first, but of even wider significance. It is that as words like persuade and still more deceit and coercion imply, the moral implications of the control of human beings are decidedly dubious. There is not time to develop either of these points as they deserve. But in a society as expressly and vociferously grounded on the ideal of freedom as ours is, it should not be necessary to elaborate this second one at great length. I am astounded at the facility with which discussions on “controlling” society and individuals pass over the essential questions of who is to do the controlling and how society is to control its controllers. In the economic field specifically I wish personally to register hearty agreement with whoever it was who made the suggestion that we ought to be subsidizing schools of resisting salesmanship instead of schools of salesmanship. And similarly in the political field. It is questionable much of the time whether our so-called criminals are either less ethical or less defiant of the actual law and constitution than are the officials supposed to safeguard the one by enforcing the other. It does not seem to me very intelligent to get all excited over developing techniques for “control” without having some advance information as to who is to use them and “on” whom they are to be used. Particularly since in view of the type of people who do get into power in democracies it seems fairly certain that the scientist himself will generally be in the group the techniques are used “on” and not the group they will be used “by”.

Irresistibly we are thrown back on the general philosophical problem already suggested but too large and too technical to go into here, the relation between controller and controlled, and between student and subject-matter. In the natural sciences it is taken for granted that these are wholly separate and directly opposed. It is “man” who studies and uses “nature!” It is a pernicious fallacy to carry over this type of thinking into the field where the student and subject-matter are of the same kind, and still more where they are identified. If the one-sided relationship is not preserved, we find ourselves committed to such absurdities as that when the scientist is experimenting with a piece of apparatus it is also in the same sense experimenting with him. The whole problem of control in society must be thought through in different terms. In any society which has aims and ideals, in any society which is not owned outright by an absolutely ruthless despot, “control” is a matter of mutual relationships, not of the one-sided character referred to by terms like control. Its members are controllers of nature and to be made in the highest degree controllers of themselves, not tools or pawns for some ruler.

The real problem of social control is the problem of securing agreement as to policy and as to the functions of individuals in promoting it where policy has to be social, and of securing the minimum of interference (“control”) for each individual in the field of what are properly his private affairs. At no important point is this problem at all similar to that confronting an engineer or any real controller. Such “control” as is legitimate in society must be “with the consent of the controlled” which makes it a categorically different phenomenon. The only exceptions admissible are the cases of individuals proven incompetent to participate in “free” society, and even those are still to be treated as far as possible as ends in themselves or ultimately perhaps as “enemies,” but in any case, never (in the modern civilized world), as means and instruments to the purposes of others, which is the position taken for granted with regard to natural objects when we talk in the scientific sense of knowledge, prediction and control.


Source: United States Department of Agriculture Graduate School. Special Lectures on Economics. Washington, D.C.: 1930. Pages 37- 45.

University of Chicago Photographic Archive, apf1-03516Image Source: , Special Collections Research Center, University of Chicago Library.

Irwin Collier

Posted by: Irwin Collier

2 thoughts on “USDA Graduate School. Frank Knight Lecture on Economics Methodology, 1930

  1. The lecture session at USDA took place a few months before the founding of the Econometric Society. E.B. Wilson was elected as one the three American council members of the Society together with Irving Fisher and Charles Roos (both mentioned in Wilson’s lecture). The other eight council members were from eight different European countries. The council compiled a list of names to be considered as charter members of the new Society. Both John Black and Frank Knight were on the list. Black got one approval (Frisch) and one disapproval, while Knight got no approval and one disapproval. Thus by the rules adhered to none of them became charter members. John Black later joined the Society as a regular member.
    One of the Europeans elected to the council was Ragnar Frisch who was invited by Irving Fisher to give a lecture to Yale students at the time of the USDA lectures. The lecture was titled “What is Economic Theory” and propagated the case for scientific economic theory, as indeed reflected in the constitution adopted by the Econometric Society at its founding.

Comments are closed.