Fritz Machlup, “The Supply of Inventors and Inventions” (1960)

This single-author conference paper in economic theory and innovation policy asks how far the supply of inventions can be analyzed like the supply of any other economic good. Machlup’s scope is narrower than technical progress: invention is separated from subsequent innovation, adoption, and imitation. The essay proceeds through three linked inquiries: the supply of inventive labor, the transformation of that labor into inventions, and the smaller supply of inventions actually put to use.

No unambiguous numerical estimates of the elasticities of supply of inventive talent and effort over substantial ranges can be derived from any data now available or likely to become available in the foreseeable future.

That warning sets the paper’s style: not empirical measurement, but disciplined theoretical ordering. Machlup first treats inventors as workers whose effort has motives only partly reducible to money.

Without trying to present an exhaustive list, we distinguish (1) inventing for fun, (2) inventing for fame, (3) inventing for serving mankind, and (4) inventing for money.

Because some invention would occur without pay, early payments to inventors may be economic rent rather than necessary inducement. But beyond that range, more inventive labor requires higher compensation, and the response is limited. Overtime and amateur tinkering cannot sustain large research programs; expansion must draw people from teaching, basic science, engineering, and other occupations.

We must not, however, neglect differences in labor quality; they are particularly important in the case of researchers and inventors.

This is one of Machlup’s central conceptual moves. “More researchers” does not necessarily mean proportionately more inventive capacity. Marginal recruits may be less suited to invention than incumbents, while higher salaries paid to attract them also raise compensation for existing researchers, producing “research workers’ rent.” He then distinguishes private money outlay from social cost: transfers to incumbent researchers are not themselves resource costs, whereas the true social cost is the value of alternative uses—especially education and basic research—foregone.

The second part asks whether inventions can be treated as outputs of a production function. Machlup recognizes the difficulty: inventions are hard to define, patents are an imperfect count, unpatented inventions may be invisible, and importance, commercial value, and technical difficulty diverge. Still, for theoretical purposes, large bundles of inventions may be “homogenized” enough to discuss whether more input yields more output. The striking formulation is that invention is a production process whose material, method, and product are all technological knowledge:

The inventor starts with technology, applies technology, and ends up with technology.

From this follows his analysis of returns. Research may initially show increasing returns through specialization and cross-fertilization, but with a given stock of knowledge and a given set of problems, adding researchers eventually produces duplication and diminishing returns. Machlup’s static marginal analysis is then paired with a dynamic qualification: discoveries can shift the whole production function by enlarging the agenda of problems.

Invention is the solution of a technological problem; but it is possible that in the course of solving a problem or as a result of solving it new problems are raised.

This distinction lets him avoid both stagnationism and technological optimism. New inventions may open new fields, but more opportunities do not necessarily make invention easier; the new problems may be harder.

The final part turns from “raw” inventions to effective ones. Machlup stresses that the path from idea to use is filtered by development costs, pilot testing, market acceptance, managerial attention, capital scarcity, and competition among alternative projects.

Not all new inventions turn out to be workable; not all that seem technically workable turn out to be commercially usable; and not all that might be commercially usable are actually used, many being rejected in favor of others that look more promising.

The paper culminates in the doctrine of the “four shrinkages”: more compensation may yield less than proportional labor; more labor may yield less than proportional capacity; more capacity may yield less than proportional raw inventions; and more raw inventions may yield less than proportional usable inventions. Its relevance for R&D policy, patent theory, and innovation economics lies in this anti-linear warning: spending more on research does not automatically produce equivalent technical advance.

These shrinkages are independent of one another; but they may add up with a vengeance.