E. 18. [Business Experience and Economists' Assumptions] [a] , [1]

We often hear that economists speculating in their cloistered seclusion hold themselves too aloof from the rough and tumble of work-a-day life for their conclusions to be of practical utility; their speculations are too theoretical, in the sense that they are based on hypotheses which have little correspondence with the facts of business.

This is closely related to a criticism concluded in more academic terms, but similar in substance. Economics, it is said, pays too little regard to psychology. It consistently uses the assumption of an economic man, omniscient and acquisitive; it ascribes all behaviour to a single motive, instead of setting out to investigate, catalogue and analyse all the varied and heterogeneous motives, which jostle together within the breast of the imperfect and wayward human beings, who set the wheels of economic activity in motion.

That there may be something in these criticisms we are driven to suspect by the results of economic studies. The best accredited masters of the subject, who may be deemed above making a slip in the logic of their arguments, appear unable to contrive to avoid differing in their conclusions. These differences indicate a difference of premises and in the last resort a difference of assumptions about how men react to various circumstances. It is beyond our powers to disentangle these assumptions and by an empirical investigation to determine which are correct?

Study of the Trade Cycle is a field fertile in disagreements of the kind described. This has suggested to a group of economists in Oxford that it might be possible by a process of intensive interrogation to ascertain how entrepreneurs do in fact react to certain specific changes in their environment and so to reach a state in which assumptions might be chosen with greater probability of their being correct, on the basis of empirical evidence. The group has had the privilege [b] of securing the presence in Oxford of a number of entrepreneurs from different kinds of industry and has subjected them to a heavy bombardment of questions; individual members of the group have also carried out similar interrogations over a wider field. [2]

I fear I shall not be able to reveal to-day any sensational results of this line of approach. For one thing my lips are for the time being sealed. Furthermore the replies have not yet been sufficiently sorted and digested. We cannot yet be sure that we shall be able to record anything more than negative results of a rather general kind. [3]

But it may be of interest to discuss certain points of method and theory to which these interrogations have continually given rise. And I propose to illustrate them by reference to one particular topic only, the treatment of overhead costs.

In the study of production the sheet-anchor of the pure theorist is the principle that the entrepreneur endeavours to arrange matters, so as to do the best for his firm. This must be conceived, of course, not only in terms of current profit, but with regard to the whole course of prospective profits. To be precise, it may be said that the entrepreneur endeavours to maximize the present value of his business, since this depends on the size of the prospective net profit from zero hour onwards.

So equipped it might be thought that we would set out on our speculations with a feeling of security; but at the very outset we are confronted with the fact that the entrepreneur is largely ignorant of the circumstances which we assume him to assess correctly for his own guidance. He is required to know the state of demand for his product now and in the future. He is also required to know his own costs. It is clear that his knowledge about the future must be strictly limited.

It might be thought that as a preliminary we could divide his decisions into two classes, (i) his decisions with regard to the installation of equipment which must be mainly governed by expectations and (ii) his decisions with regard to current output, prima facie affected by current conditions only. Can we not obtain a fairly precise theory concerning the second set of decisions?

Unhappily even these are seriously affected by considerations about the future. He may charge a price which having regard to the current state of demand is the right one. But if in so doing he loses custom, we must deduct from his current receipts the present value of the goodwill lost in order to determine his true current net revenue. But how can the present value of the goodwill lost be assessed unless correct pre-vision is possible? [4]

Again even the estimate of his own current cost of production, which, if anything, appears on the face of it capable of accurate assessment, is infected by the virus of doubt about the future course of events. Mr. Keynes has well shown, by means of his concept of User Cost, how the true cost of the wear and tear of equipment can only be properly gauged if it is known whether and to what extent the equipment will be of value in the future. If there will be no future demand for its product, its present wear and tear should be set down as zero; but not otherwise. [5] I should like to refer in this connexion to an article in the Economic Journal (September, 1936) on Prime and Supplementary Costs by Mr. G. D. A. MacDougall, [6] which investigates the problem much more elaborately and brings out the full and horrid complexity of the correct definition of such an apparently simple quantity as current cost of production.

Nor, even if we could isolate the present from the infection of uncertainties regarding the future, would full knowledge reign. The entrepreneur is required to have knowledge of the present state of demand. A state of demand can only be represented by a schedule relating various prices to the amounts which the market will absorb at each, and here again we find ourselves plunged into a dense fog. Information on this point can only be gained by trial and error conducted in an ever-changing environment. Our questions indicated that entrepreneurs are often only able to make the vaguest conjectures regarding the current elasticity of demand.

A decision to be fully rational requires complete knowledge of all the relevant data. This is scarcely ever obtainable. By far the greater part of economic decisions are thus to some extent irrational.

Most entrepreneurs appear ready to admit and even to emphasize and exaggerate their ignorance of the relevant data. But large differences may be found with regard to their confidence in the validity of their own decisions. This no doubt depends largely on temperament. Some readily admit that they act on guess-work with little to go on and that their decisions are often as likely to be wrong as right. Others incline to the view that despite their ignorance of data expressible [c] in quantitative terms they arrive by the process of judgement at conclusions which are approximately correct.

Now it is no doubt quite possible to arrive at a correct conclusion which has quantitative implications as regards its premises without being explicitly aware of those quantitative implications. This is the general rule in skilled motions of the body. From the conclusion it should be possible to argue back to the quantitative premises implied. If 1/- and not 11d or 1/1d is the right price to charge, this implies that the demand has a certain specific elasticity etc. Such complete ignorance, however, is often affirmed with regard to these premises by those who rely on judgement that some scepticism with regard to the validity of their judgements may be pardonable.

While ignorance is widely admitted, it is usually possible to set limits to it. Thus there may be certainty based on good reasoning that it would be wrong to charge double or half the price chosen. Thus there may be a determinate range of ignorance.

The economist who argues from the assumption of full knowledge might justify his procedure in this way. If decisions are distributed at random within the range of ignorance, then, if there are a sufficiently large number of agents operating, the net result would be the same as if perfect knowledge obtained. This would give a reason, additional to those associated with the theory of imperfect competition, why traditional competitive analysis is becoming less applicable in the real world. The sphere in which there are a sufficiently large number of operators to justify the assumption of knowledge by the theory of probability is becoming more restricted.

Now if we are to make progress away from the assumption of omniscience, it is necessary to be able to make some classification of the factors affecting decisions within the range of ignorance. The greater the confidence the entrepreneur has in his judgement the less easy it is to do this. When pressed for an explanation he has nothing to add. But if he is more diffident and perhaps more realistic, he may be willing to admit that he is governed by this and that consideration, albeit not strictly relevant.

An investigation into factors governing action within the range of ignorance might be divided into those which affect judgement consciously and those which affect it un-consciously. The latter is perhaps the field in which the economist might in principle hope to learn most from the psychologist. The so-called psychological theory of the trade cycle, with its alternating waves of optimism and pessimism, is well known, though its psychological basis is something of an improvisation on the part of economists. [7] I have never been altogether able to understand it. It is usually assumed that optimism leads entrepreneurs to produce more than in view of the facts they ought to. Why should it not equally well lead them to ask too high a price and so produce too little? In the whole field in which goods are disposed of at prices stated in a catalogue and also in that at which orders are obtained by competitive bidding, the effect of optimism seems doubtful. Optimistic opinion may take the form of a belief that the demand is more elastic than it is and exaggerate the possibility of expanding the market by price reduction; or it may take the form of a belief that the demand is inelastic and that if you ask a high price, you will get it. The psychological theory seems to require an additional and surely unwarrantable assumption that optimism is tantamount to a firm faith in elasticity and pessimism to a firm faith in inelasticity.

The method of questionnaire [d] is clearly not very appropriate to unravelling un-conscious factors. [8] It is possible that the lessons of experience may bring up into the entrepreneur's consciousness for condemnation, factors which had been unconsciously guiding him in the past. But the lessons of experience may remain in the un-consciousness also. Moreover in practical life the urgency of present affairs makes the past recede very quickly. In many entrepreneurial minds the problems and struggles of 1931 have already taken their place with 1066 and all that.

What of the conscious factors? Though I do not propose to present any definite findings, it appears that the treatment of overhead costs illustrates the kind of factors that may operate within the range of irrational choice.

Strictly speaking it is a commonplace among economists that overhead costs should play no direct part in the fixing of a price. 1 Entrepreneurs should not, save in a case to be mentioned, quote a price below the direct or prime cost of production. They should charge such a price above the prime cost as will having regard (i) to the current elasticity of demand and (ii) to repercussions of their present price on the future demand for their goods maximize their profit present and future, in such wise as to maximize the present value of their business. They may go below the prime cost if they estimate that the present value of the loss of future profit resulting from the loss of future customers due to their charging a prime cost price rather than a lower one exceeds the present loss constituted by the difference between prime cost and the said lower price.

This is the strict rule of theory. Unhappily the two facts enumerated, namely the present elasticity of demand and the future loss of profit due to charging a particular price now are precisely those about which entrepreneurs are largely in ignorance. Consequently they do not know what is the right price to charge by this criterion.

On the other hand, by contrast with this vague and nebulous factor of demand elasticity, they may have or anyhow seem to have fairly precise information with regard to their overhead costs. These are, it is true, surrounded with some doubt and ambiguity. But this doubt is less or may appear on first inspection to be less than the doubt attending an estimate of demand elasticity present and future. Cannot this knowledge concerning overhead costs be put to good use? Can it not be brought in to fill in the gap due to their ignorance of the state and trend of demand? Specifically, why should they not make it their rule to charge such a price in excess of their prime cost as may be expected to cover overhead costs?

I do not suggest that entrepreneurs invariably or even generally charge a price equal to the prime plus the computed overhead cost. Evidence abounds that this is not so. But on the other hand there is considerable evidence that they do attach some weight to the computed overheads in forming their price policy. Practice differs considerably. Some have an almost completely rigid rule that they will not go below the price which covers prime plus overhead cost. Others pay scant attention to overheads; but it appears to be rare for them to pay no attention at all.

This may be of considerable significance for theory. For it suggests the possibility of a quite specific bias within the range of ignorance. If the range of ignorance for a group of producers so far as concerns what is best from the demand point of view lies from 1s to 2s and their computed cost including overheads is 1/9, it is probable that the mean price charged will be I do not say 1/9 but slightly higher than 1/6.

Some attention to overheads may indeed be justified on rational grounds. The argument would run as follows. We do not in the least know how the market will react to various prices; all that is a matter of the barest conjecture. But we do know this. In the long run overheads must be covered if the firm is to remain in business; and in the long run also the firm which charges a price in excess of cost including overheads is likely to lose ground to its competitors. There is also a less rational justification. In a firm which has to quote a number of prices, some routine method of reaching a price may be necessary. To make a fresh decision based on a complete survey of the market would be too cumbersome and too costly of managerial intelligence. Some rule of thumb may be required. The costing principle provides such a rule.

The bias in favour of covering overheads is clearly relevant to the study of the sequence of events in the Trade Cycle. During the depression the margin over prime costs required to cover overheads becomes greater at a time when demand considerations suggest a smaller margin. Thus the bias if operative would hold prices at a higher level and thus reduce production to a lower level than all the relevant circumstances require.

It must not be inferred that attention to overheads tends to intensify the amplitude of the cycle itself. That would constitute a fallacious inference from the part to the whole. Indeed if a theory of the cycle of the type propounded by Mr. Keynes is correct, it would work the other way. On this theory the recession proceeds until the volume of voluntary saving is reduced to the level required by the investment opportunities then available. Any deviation of output from the best possible level tends to reduce profits or increase losses; but the quicker the accretion of losses as production recedes, the smaller has the recession of production to be to reduce saving to the necessary level. [10] On this view more refined methods of adapting output to demand would intensify the amplitude of the cycle, just as a more [e] accurate cutting of the coat according to their cloth by consumers is thought to make the depression worse.

To profess regard for overheads is not a final explanation of policy. It merely leads to the question of how overheads are computed, and in particular to the volume of turnover presupposed in fixing the margin required to cover overheads.

Of firms expressing allegiance to the overheads principle a surprisingly large number use very crude methods of computation. This may well be because the attempt to use more refined methods demonstrates the un-soundness of the system and leads to its abandonment.

Systems of costing for overheads may be divided into two broad classes, those using a flat rate and those based on forecasting. What I call the flat rate may take one of two forms. 1. Some fixed margin such as 20% may be added to direct costs. This is based on past experience and only revised at long intervals. 2. Overheads may be accurately computed annually or at shorter intervals, but the required margin on oncost is computed on the basis that the plant is worked to full capacity or to some specified percentage, such as 80, of capacity. [11] These methods have the cardinal feature in common, that there is no tendency for the margin to swell as output recedes.

They are clearly very imperfectly designed to fulfil their alleged purpose, namely that overheads should be covered. They both go some way in tempering the price to the market and this may account for their survival.

The alternative method is to attempt some kind of forecast. This is more refined. But even in these cases the firms usually make their forecasts by some very crude method. The commonest is to assume that the output of the coming year will be the same as that of the year preceding [f] . This method would tend to pile on the oncost during the depression but with the lag of a year. A few very enlightened firms may attempt some more complicated form of extrapolation; but this is very rare. This is not to say that many firms do not have a department which is thinking ahead and making shrewd estimates; but the results of such estimates do not usually filter through the routine work of costing.

Four points seem to become clear.

1. The same instinct, which makes a firm pay attention to overhead cost in fixing prices, namely a preference for the terra firma of ascertainable fact to the realm of vague conjecture, would incline it to the crude method of a flat rate rather than one based on forecasting. This may be one reason for the widespread survival of the crude method.

2. The crude method is likely to be retained, if it achieves its purpose fairly well, namely if it does get overheads covered. This will occur if the fluctuation in demand is moderate.

3. If the fluctuation is great, no method can succeed in this object. The same cause which makes the crude method unsatisfactory, tends to discredit allegiance to the cost principle entirely. A refined method will pile on the oncost in the vain endeavour to cover overheads, but reference to demand would prescribe a reduction of oncost in order to secure the maximum total allocation to overheads. Thus if fluctuation is great a reference to overheads would lead price quotations in the opposite direction to that required; in such circumstances the reference tends to be abandoned altogether as in the cotton industry in recent years. Yet even there the entrepreneurial instinct for regarding overheads is so great, that there is a firm belief that their abandonment is abnormal and even immoral, and recent price agreements contain a proviso enforcing regard for them.

4. In order to obtain an accurate assessment of oncost, it is necessary to have precisely that information, the lack of which is the primary reason for regarding overheads at all. No true estimate for oncost can be made without an estimate of prospective turnover; turnover depends in part upon the price; but the price itself is what has to be determined. Thus costing procedure, to be perfectly scientific, would have to form an estimate of the volume of demand at each possible price, estimate the oncost for each such volume, and thus ascertain the price most nearly equal to the prime cost plus oncost appropriate to the output which could be marketed at that price. True costing is in fact impossible without an estimate of the whole demand schedule. But if it is possible to make such an estimate, the price may be fixed directly by reference to it and the whole ritual of arriving at a true oncost becomes completely unnecessary [g] .

The matter may be summarized more briefly thus:--

1. If the crude method is unsuccessful [h] in getting overheads covered, no method is likely to be successful.

2. A forecasting method while paying more attention to demand than the crude method tempers the price to the demand less than the crude method.

3. If it is desired to avoid the uncharted seas of conjecture only the crude method is available.

4. If it is not considered necessary to avoid these uncharted seas, all reference to overheads is otiose.

In the actual world a large variety of methods exist side by side and varying degrees of attention are paid to the results of the computation in price formation. I hazard the conjecture that the more refined the method the less attention is paid to the result.

I have the impression that some kind of attention to overheads is very widespread, sufficiently so indeed to have a marked effect on global output and the course of the trade cycle.

It may surprize some that many firms have sufficient latitude in their price quotations to allow this. A word of explanation is necessary. It often happens that price is related to cost in a slightly more complicated way than that suggested by the foregoing. The price may be fairly rigidly fixed by competition or market custom; the costing is in those cases often used to determine the quality of the commodity which the firm offers at the price. But to reduce the quality offered at a given price in deference to the cost criterion is similar in principle to raising the price while maintaining the quality intact. The argument with regard to the effect of such policy on the course of the trade cycle would be similar to that concerning open price adjustment.

I have used the treatment of overheads to illustrate how a rule of thumb, containing a specific bias, may be brought in by the entrepreneur to fill the gap due to his own un-avoidable ignorance. It is only intended as an illustration. There are many other matters, that will readily occur to your mind, extension of plant, depreciation policy, reserve policy, to give but a few examples, in which there is the same ignorance of all the relevant data and the same tendency to adhere to some criterion, tested perhaps by experience, but not strictly relevant or rational.

Tradition probably plays a considerable part in the formation and conservation of these rules of thumb. If some arbitrary procedure has stood the test of time, it tends to be retained. Quasi-ethical considerations, what is considered "fair" treatment of customers, of employers, of shareholders may also come in; [12] all such considerations should be carefully sifted for a possible factor biasing random choice within a field of ignorance.

These reflexions suggest a different interpretation of the process by which competition leads to the survival of the most efficient firms. We are apt to think of the more efficient producers as being more intelligent, industrious, punctual etc. But it may be that industrial evolution proceeds more on the lines of evolution in nature. New businesses arise and as a matter of practical expediency are bound to adopt some routine procedure with regard to many matters of policy within the field of ignorance. In so far as they lie within this range or--and we must remember that it is often great--, no one procedure can be regarded as more intelligent than any other. Yet in fact one may give results lying, on the average over a term of years, nearer to the optimum than another. Competition will weed out the procedures less well adapted to the environment.

The adoption of new procedures is analogous to mutations in the theory of natural selection. [13] The more frequent the occurrence of new mutations, the greater the opportunity for industrial adaptation. The greatest opportunity for new mutations occurs on the birth of new firms. Adaptation to environment would in our analogy consist in the spread of a procedure which in fact resulted in the closer approximation to the level of output required to maximize profit. Such adaptation is more likely to occur if mutations are frequent. Thus it might be the case that in circumstances of a very high birth-rate of firms, the assumption of profit maximization would be a closer approximation to the truth, than in those of a much lower birth rate. Advocates of universal cartelization should bear this in mind. It may be long before ignorance can be so narrowly circumscribed that intelligence can take over the part performed by the healthy process of natural selection.

  1. 1. Although this essay bears no title, it seems to correspond to the paper read on 3 September 1937 before Section F of the British Association for the Advancement of Science. It largely matches the abstract published in the British Association's Report of the Annual Meeting, 1937 (London: Office of the British Association, 1937, pp. 56-58), and is the only extant document relating to the activities of the Oxford Economists' Research Group written to be read before an audience. Harrod commented on the meeting in two letters to his future wife: see letters 692 R and 693 R.

    Harrod was asked by the OERG to read a paper to the 1937 meeting of the British Association on the programme of the group (minutes of the Committee Meeting of 3 February [1937], in ABP 45). This documents develops some of the ideas presented by Harrod in his "Notes on Interviews with Entrepreneurs" (here reproduced as essay 17 ).

    For Harrod's own brief impressions on the paper as read before the meeting see letters 692 R and 693 R.

    2. On the OERG method of inquiry by means of questionnaires, see note 5 to essay 17 . In addition to the interviews taking place in Oxford before the whole group, individual members visited firms around the country. Harrod, for instance, visited some entrepreneurs in Leicester in 1938: see note 1 to letter 763 R.

    3. The entrepreneurs' recorded replies showed wide disagreement on most subjects, with two exceptions only. First, there was unanimity in the negative reply to the question whether the bank rate is an important consideration in the decision regarding expansion or contraction of activity, and only one exception to the chorus of denial of any connection between increase of stocks and a fall in the bank rate. All but two entrepreneurs also denied the influence of the long-term rate of interest. Secondly, the answers to the group of questions on costs, allocation of overhead and price policy showed a remarkable consistency: the firms that managed to "shelter themselves from the full force of competition [...] incline on the whole to base price policy on costs"; "those which have not [...] take what they can get" (OERG, "Analysis of reports of interviews with business men. February-October 1936", 44 pages mimeograph, undated, not signed nor initialled, in ABP 171). Naturally, the group focused its attention on these subjects: prices and costs were discussed by Harrod, Charles Hitch and Robert Hall (see note 2 to essay 17 ). The other set of questions was taken up by James Meade in September 1937: "Analysis of Information received on factors determining the volume of investment" (in ABP 46, 173 and 630, and HCN 4.30.1--the latter with annotations in Harrod's hand).

    4. Harrod's discussion of the goodwill factor in "Notes on Interviews with Entrepreneurs" was more detailed (essay 17 , in particular [jump to page] , [jump to page] and [jump to page] ). He may have abandoned this line of analysis as a consequence of the anonymous comment on his "Note" reported in note 16 to essay 17 .

    5. Keynes, The General Theory (1936), pp. 66-73, in particular § 2. The concept was discussed with Harrod in correspondence at the proof stage: see letters 458 , 463 , 464 , 465 , and 475 . See also Harrod's "Notes on Interviews with Entrepreneurs": essay 17 , [jump to page] .

    6. G. D. A. MacDougall, "The Definition of Prime and Supplementary Costs" (1936), pp. 443-61. Harrod discussed MacDougall's article with Keynes in July 1936: see letter 574 .

    7. Harrod probably referred to the approach developed by A. C. Pigou in Industrial Fluctuations (1927), which he criticized in "Doctrines of Imperfect Competition" (Harrod 1934:3 , pp. 465-70).

    8. On the method of inquiry by means of questionnaires, see note 5 to essay 17 .

    9. Refers to § 1 of the section on "Volume of Output" of Harrod's "Notes on Interviews with Entrepreneurs" (essay 17 , [jump to page] ), where this point is discussed more in detail.

    10. In Harrod's "Notes on Interviews with Entrepreneurs" the point was made without explicit reference to Keynes, but referring instead of Harrod's own argument regarding the shift away from profits in the downswing: see essay 17 , [jump to page] .

    11. The examples are drawn from Harrod's summary of replies of entrepreneurs, reproduced in note 10 to essay 17 .

    12. The influence of quasi-ethical factors as a basing element in pricing policy was originally suggested by Henderson (see note 2 to essay 17 ). Later Harrod took up the subject of entrepreneurs' morality in "Price and Cost in Entrepreneurs' Policy" ( 1939:10 ), pp. 8-11, and discussed it with H. F. Scott-Stokes and B. Ellinger, two of the entrepreneurs interviewed by the group (see letter 917 R, and in particular note 1 for context).

    13. The analogy of the natural selection was also referred to in Harrod's "Notes on Interviews with Entrepreneurs": essay 17 , [jump to page] .

    1. a. AD in Harrod's hand, not signed nor initialled, 11 pages on 11 leaves, in HCN 5.13.5.

      b. Ms: «priviledge».

      c. Ms: «express-able».

      d. Ms: «questionaire».

      e. Ms: «more more».

      f. Ms: «preceeding».

      g. Ms: «un-necessary».

      h. Ms: «un-successful».


1. sub pp. 9-10 of typescript [Harrod's note in the margin]. [9]


Welcome page

top of page

Return to index of this section

Go to previous page

Go to next page