be determined independently of the extent of use that is made of it; and the extent of use depends, among other things, on the royalty that is charged. In brief, it is not the usefulness which can determine the royalty rate; rather, the royalty rate will determine the use, and with it the usefulness, of the licensed invention, given the known substitute technologies and the demand for the product.250

The fuller the utilization of the invention the greater, of course, is its total usefulness to the consuming public; but the smaller also is its "marginal" usefulness. The more fully the invention is exploited, the lower will be the prices paid by the consumers for the final prodThe height of the royalty rate will determine how far the producers can go in the utilization of the invention. As long as a payment for royalties can be squeezed out of the pockets of the buying public, one could go still further in satisfying its demand. Since the marginal cost of using any existing invention is zero, it follows that only when its marginal utility is zero will its total usefulness to society be maximized.251

Can this total social usefulness of an invention, whether it is used "fully" or "with restraint," be estimated? Certainly not by what is paid for the use of the invention. There is some possibility of estimating in money terms the social benefit rendered by a cost-saving invention. If, thanks to such an invention, fewer productive resources are needed than before to produce a given quantity of product of given quality; and if the productive resources economized by using the new process can be employed for producing either more of the same good or more of other goods, the Nation's total output will be greater. This increase in national product due to the invention can be estimated by the competitive prices of the resources economized in the production of the original output. For example, if an invention permits an annual net saving of $1 million worth of labor and material, and if there are uses for the released labor and material, one is safe in estimating that the invention has a social value of $1 million per year. There is little possibility, however, of estimating the social benefit of a qualityimproving invention, and almost no possibility in the case of inventions of new products. That people are better off with the new products than they had been with what they used to buy, is generally assumed provided their choices are free. But any numerical index for translating a change in the composition of output into an increase in output would be quite arbitrary.

In any event, even if there existed ways of estimating the social value of new inventions, how is this connected with the issues with which we are dealing? Let us recall that we are not talking now about the value of patents, nor about the social value of the patent system, but rather about the social value of inventions. Again several different questions must here be distinguished: the social value of a particular invention; the social value of the annual crop of inventions,

250 In this argument the royalty rate was the independent variable and the quantity produced (1. e., the degree of utilization of the invention) was the dependent variable. One can turn it around and make the quantity produced the independent variable, and the royalty rate the dependent one. This would be like asking how much the licensee could afford to pay for the permission to use the invention for a certain volume of output. There is nothing wrong with a statement that the usefulness of an invention to a licensee is reflected in the royalty rate he would be willing to pay for a fixed volume of output rather than do without a license. This would be equivalent to the statement that the usefulness of an invention to a licensee is reflected in the volume of production for which he would use it at a fixed royalty rate per unit of output. In both ways of looking at the problem the volume of output (or degree of using the invention) is crucial and must not be disregarded.

251 Adepts of the differential calculus will easily recognize that total utility is a maximum when the first differential coefficient-marginal utility-is zero.

patented or unpatented; the social value of the annual crop of patented inventions; and, lastly, the social value of the annual crop of patentgenerated inventions, that is, of inventions that would not have been made or developed had it not been for the incentives afforded by the patent system. This increment of invention that is attributable to the operation of the patent system is probably of relevance to an evaluation of the patent system as a whole. But there is yet another magnitude, perhaps even more interesting for the problems before us: the (positive or negative) increment of invention that is attributable to certain changes in the patent system. The possibility of analyzing these two increments will occupy us in the next sections.

J. THE COST AND VALUE OF ADDITIONAL INVENTIONS

The analysis of the "increment of invention" attributable to the operation of the patent system, or to certain changes in the patent system, can only be highly speculative, because no experimental tests can be devised to isolate the effects of patent protection from all other changes that are going on in the economy.

May we "dream up" some experimental testing of the differences between a world with patents and one without patents? Let us duplicate our world, so that we have two worlds identical in every respect, except that one shall have a patent system and the other shall not; and then let us observe, for 50 years or so, these identical twin worlds and see what happens. And let us also have identical twin worlds of the years 1700, 1750, 1800, 1850, and 1900, one of the twins always with and the other without a patent system. It is conceivable that such "experiments" would yield trustworthy results, especially if we were able to repeat them and control some of the other factors that might make a difference to the rate of technological progress. It is also conceivable that the findings would be somewhat inconsistent: For example, the worlds of 1700 and 1750 might show superior progress in the specimen equipped with patent systems; the worlds of 1800 might show no differences in the rates of progress; and the worlds of more recent vintage might show faster progress in the specimen without patents. Such findings would be in accord with the hunches of some writers of the late 19th century, who hypothesized that the patent system may have been useful in kindling the spirit of inventive ambition, but is unnecessary or harmful once industrial inventiveness is sufficiently developed.252 Yet there is no use imagining the findings of the imaginary experiments. There are no real experiments that can answer our questions and we have to fall back on speculative analysis, on inferring conclusions from assumptions which, on the basis of common experience ("casual empiricism"), seem to be the most plausible.

One may be fussy and contend that it makes no sense to speak of an "increment of invention" (attributable to the patent system) because inventions can be neither counted nor weighed nor measured in any practical way. Perfectly true. Inventions can often be subdivided or fused, and hence counting is arbitrary; and even if one

252 "The wisdom of our ancestors is not discredited when, now that circumstances have completely changed, we abandon a system of restraints that is no longer tenable. British manufacturers have outgrown the confinement and trammels of the nursery and go-carts, and demand freedom of action and fuller scope." Robert Andrew Macfie, The Patent Question under Free Trade (London: second edition, 1864), p. iv. "In early stages of industrial development patent protection may have been beneficial. Not in the present state of the economy." Hermann Rentzsch, "Patentwesen", Handwörterbuch der Volkswirtschaft (Leipzig: 1866), p. 634. Similarly, Sir Roundell Palmer (in the House of Commons] as reported in Westminster Review,new series, vol. XXXVI (July 1869) p. 125.

agrees on some system of counting, one must realize that there are highly important and altogether nugatory inventions, and that it would be silly to give them equal weight. Yet, when all this is said and done, one will still have to concede that it is not meaningless to say that some times have been more productive of new inventions than others, and that some conditions may be more conducive to inventive success than others; and what can this mean if it does not mean "more" inventions? If more people are put to work on industrial research and development, more inventions, important as well as trifling ones, will be produced. The exact meaning of the "more"of the increment-may be in doubt. But we need not be so fussy, and may be satisfied with something less exact. Incidentally, since we are going to use the concept of the "quantity" or "amount" of invention only in a speculative analysis, we may proceed as if we were able to give an exact meaning to the concept.

The bulk of technological advances, especially the millions of small improvements in production techniques which probably account for a large part of the increases in labor productivity, have nothing to do with patent protection. This can probably be tested by examining the types of technological change made over recent years in many different industries. 253 Thus, only some part (of unknown size) of all increases in productivity is derived from patented inventions. Of these inventions, some might never come into being without the patent incentive; others might come later; and the rest might come in any case and at the same time, with or without patents. This means that the patent system is not to be credited with all patented technology, but only with that technology obtained "only with patents" and that obtained "earlier with patents."

Granted, that there results an increment in national product attributable to inventions that are generated, or whose application is accelerated, by the patent incentive. Against this, however, must be set the reduction in national product that is attributable to restriction in the use of those inventions which are patented but which would have appeared at the same time without patent incentive and would have been free for unrestricted use by anybody. The restrictive effects of the patent system are not confined to those immanent, or inherent, in the exercise of the patent monopoly; that is, to the relative underutilization of the patented inventions. Besides these "immanent" restrictions there may be "transcendent" restrictions associated with the increased strength of the patentees' general monopoly control in their markets.254 Account should also be taken of possible

Most

Every plant superintendent introduces hundreds of small technical improvements every year. of these are quite trivial-relocating some machines; changing transmissions, conveyors, pipelines; readjusting temperature, light, pressure, rotations, water contents; using more suitable materials, fuels; avoiding waste; altering sequences of operations; rescheduling of repair and maintenance work-perhaps nowhere recorded, but they may add up to a substantial total effect upon productivity.

24 The terms "immanent" and "transcendent" restrictions are introduced here in recognition of the pilor rights which patent lawyers have in better-sounding phrases such as restrictions "inherent in the patent grant" and restrictions which are "unauthorized extensions of the monopoly" or go "beyond the scope of the patent monopoly." "Immanent" restrictions, in an economic sense, are not coextensive with "inherent" ones, and "transcendent" not with those going "beyond the scope of the monopoly grant." For example, a patentee using his control to compel his licensees-lessees to buy from him an unpatented material for use on the patented machine is illegally going beyond the scope of his patent monopoly; yet the imposed restrictions are still immanent in the exercise of his monopoly in the economic sense used here, because the (unlawful) actions restrict only the use of the patented technology (and the patentee could achieve the same effects by setting royalty rates in proportion to the amount of material used). On the other hand, the use of the general market power gained by the patentee as a result of his patent position may lead to transcendent restrictions, that is, limitations on the output of different commodities and, hence, in the use of different technologies; these restrictions could not be identified by the law as extensions of the patent monopoly. The choice of this new economic terminology, though it avoids infringement of prior rights in "words of art" used by lawyers, may involve an encroachment on the domain of Kantian philosophy-but philosophers take such matters philosophically.

"obstructions and encumbrances" which patents may put in the way of others, to wit, potential inventors and innovators, and which keep them from engaging in industrial research in certain directions, from working on ideas the development of which seems blocked, or from undertaking innovations which, though not really infringing any patents, might incite harassing litigation. Thus not only is the use of existing technology restricted-this, to some extent, is intended by the patent system-but possible developments of new technology may be interfered with by existing patents.

Three other cost items also have to be taken into account: (1) the cost of operating the patent system, which means chiefly the input of administrative, legal, technical, and clerical ability in government, industry, and law offices; (2) the cost of inventing, which is primarily the use for industrial research and development work of scientific and engineering personnel withheld from other activities; and (3) the cost of innovating, which consists of faster obsolescence of capital goods and of losses due to more frequent transfers of human and material

resources.

Thus, the benefits derived from the patent system consist in the increase in national product attributable to technological innovations which are "generated" by the system in the sense that they would not come into being without patent incentive or would arise only at a later time. The costs, or negative items to be set against the benefits, can be organized under six heading: (1) the operating cost of the patent system, (2) the cost of inventing, (3) the cost of innovating, (4) the cost of immanent restrictions in the use of patented inventions, (5) the cost of transcendent restrictions upon production as a result of general monopoly control strengthened through patent positions, and (6) the cost of obstructions and encumbrances to potential inventors and innovators. Most patent experts take it for granted that the "generating capacity" of the system is great, and that its restricting and obstructing effects, as well as the other cost elements, are neg ligible. Of course, no ready means of measuring the positive and negative effects are available, but one should expect at least some theoretical analysis to precede pronouncements on the largeness of net benefits.

To illustrate, one of the six cost items may be singled out at this point because it has a bearing on the most essential arguments: the cost of inventing. One must assume that beyond a certain volume of inventive activity the cost of inventions increases rapidly, because the "production of inventions" is liable to become subject to drastically diminishing (if not zero) returns 255 and, moreover, the supply of inventive talent is, beyond a point, highly inelastic. If inventive

255 "Diminishing returns" in the sense used here mean that the "output" increases at a smaller proportion than the "input," so that the cost per unit of output increases. There is usually a phase of "increasing returns"-where output increases proportionally faster than input-before diminishing returns set in. It is quite possible, therefore, that a nation can still increase the production of inventions at increasing returns: that, for example, a 10-percent increase in the inventive talent employed for industrial research and development will produce a 20-percent increase in the flow of inventions. Moreover, it is possible that inventive activity at one time goes on under drastically diminishing returns, but then an important scientific discovery suddenly opens up such a wealth of problems of practical application that the production of inventions moves into another phase of increasing returns.

Even if a nation has allocated enough resources to the production of inventions to have pushed it far into the range of diminishing returns, this need not mean that too many resources have been so allocated. Indeed, economists can explain why production is most efficient under diminishing returns. Thus, it is not to charge wastefulness if it is said that the production of inventions is subject to diminishing returns. It may be well worth trying for a 2-percent increase in the flow of inventions at the expense of a 10-percent increase in the employment of research personnel. All that the possibility of "drastically diminishing returns" should mean to us is that we ought to watch the cost and not be deluded by the false hope that a given percentage increase in research staff will always yield the same percentage increase in inventions.

activity has been pushed that far, it may mean that a further increase in the research staffs of all companies and institutes by, say, 10 percent may yield an increase in new workable inventions by only, say, 1 or 2 percent; and the increase in demand for research personnel may boost the research payroll by, say, 30 percent for an increase in the work force of only 10 percent. Thus, a relatively large outlay may be needed to produce a relatively small increase in the production of inventions. In addition, inventions are subject also to rapidly diminishing utility, because a greater volume of inventions will ordinarily include a higher proportion of multiple inventions, of substitute inventions, of process inventions for the production of products simultaneously made obsolete by new product inventions, etc.-and because the number of inventions that can actually be put to use is limited by the available supply of productive resources and capital, which will compel a more stringent selection from the inventions supplied.

The double action of diminishing returns and diminishing utility is particularly important in evaluating the social desirability of changes in the patent law, especially in the scope, strength, and duration of patent protection. It is sometimes assumed that the "best" patent law is the one that gives patent applicants the biggest chance of obtaining the safest protection for the longest time. This assumption is made without any attempt to examine how effective an extension of the scope, strength, or duration of patent protection is likely to be in inducing the desired technological advance. 256 Yet, such an examination is essential, and to make it the following questions must be answered:

(1) How much would a small increase in the length, strength, or scope of the patent monopoly increase the profit anticipations of those who invest in research, development, and actual innovation?

(2) How much would this increase in profit anticipations raise, at effective interest rates and risk allowances, the present value of the expected returns?

(3) How much would this increase in the present value of the expected returns increase the amounts of funds currently invested in research and development?

(4) How much would this increase in current investment in research and development increase the amounts of productive resources, chiefly human resources, allocated to research and development work?

(5) How much would this increase in the current input of productive resources for research and development increase the output of novel and useful technological ideas?

(6) How much would this increase in the output of technological ideas increase the rate of actual execution of innovations in production? (7) How much would this increase in the actual rate of innovation in production raise the productivity of productive resources?

(8) How much would this increase in productivity of labor, land, and capital goods increase the national product?

(9) To what extent would this increase in national product be offset by the decrease in national product that would result from the output restrictions inherent in the extension of the patent monopoly?

255 A few writers have stressed the effects of patent protection upon the rate of investment (and employment) more than the effects upon the rate of invention. If one assumes that there is no scarcity of investment opportunities, one may expediently restrict the analysis to the effects on technological progress, which is in conformance with traditional patent theory.