Oskar Morgenstern’s periodical article examines what would be required to compute an economic program rapidly enough, and accurately enough, to guide public decisions. Beginning with military mobilization, it moves through economic interdependence, mathematical modeling, aggregation, and resource constraints before illustrating interindustry calculations with input-output tables. Its central argument is that neither administrative experience nor isolated engineering estimates can establish the consequences of a large program. Those consequences depend on relations across the economy, and determining whether a program is possible remains distinct from determining whether it is best.
The opening example makes this problem concrete. The armed services separately specify their requirements, but ships, tanks, and airplanes compete for materials and labor while depending on transportation that must itself remain adequately supplied. Diverting steel from railroads could undermine the movements needed to execute the military program. Calculation must therefore establish compatibility among demands, the availability of manpower, and the necessary reduction of civilian activity. Speed matters because strategic circumstances may change before the calculations are finished. Morgenstern compares this predicament to weather forecasting whose computations arrive after the weather itself.
Moreover, the only acceptable solution would be a quantitative one, which would show us the actual magnitudes of the expected shifts in activity of the various industries.
Quantification here means tracing consequences, not merely attaching totals to a list of requirements. The desired program should meet military needs as nearly as possible while minimizing disruption to ordinary economic functioning. The same problem arises in peacetime when health provision, old-age security, and public construction are proposed simultaneously. Individually desirable measures may obstruct one another; their compatibility cannot be judged independently of their shared economic requirements.
Morgenstern distinguishes two forms of interdependence. Following Quesnay’s account of economic flows, he describes commodities and services moving in one direction and money payments in the other. Changes in production, monetary supply, or spending affect participants beyond their control. More subtly, people act in anticipation of others’ conduct: this strategic dependence requires game theory rather than mechanical analogy. Technological dependence concerns the materials and services required for production, the limited possibilities of substitution, and competition for common facilities. Automobile production and agriculture, for example, become linked through their demands on railroad capacity.
Because of the intricacies and the often unexpected character of these interrelations, no intuitive approach based on mere "experience" is satisfactory.
This rejection of intuition motivates mathematical modeling, but Morgenstern does not equate a model with a complete economic description. Individual transactions are too numerous to record and compute, even with electronic machines. Information must be compressed into industries or larger groups. Aggregation makes calculation possible while sacrificing distinctions: transactions between firms may appear as an industry buying from itself because both firms belong to the same statistical category. The argument thus combines the necessity of formal representation with recognition of its informational limits.
The next conceptual move separates constraints, feasibility, and optimality. Steel capacity cannot expand instantaneously, and labor cannot be increased without considering employment, skills, working hours, retirement, or transfers from other production. Wartime trial and error exposed the costs of discovering these limits through practice. Yet the decisive constraints are not simply engineering or physiological limits; they also arise from relationships among economic activities. A technically achievable expansion in one sector may conflict with requirements elsewhere.
If a program can be shown, by computations, to be feasible, this merely means that one solution--not necessarily the best solution--has been found.
Feasibility establishes that a program fits the relevant restraints; optimality asks whether further improvements remain possible. Morgenstern connects this distinction to business decisions: earning a profit does not demonstrate that the maximum attainable profit has been secured. National programs are harder still. Although logistics presents related feasibility problems, the economy lacks centralized control over its millions of decisions. Prices, expected profits, and costs mediate behavior, so economic computation cannot be reduced to arranging material deliveries.
The concluding discussion measures these ambitions against contemporary capabilities. The Bureau of Labor Statistics’ collection of 1947 input-output data, following Leontief’s interindustry formulation, promises a table of 150 industries and 22,500 transactions. Yet availability is expected only in 1951, with interpretation taking additional time. Better information may already describe an economy that has changed.
However, the point is that we do not have, at present, powerful methods that allow the rapid and accurate computation of economic programs.
Two tables nevertheless demonstrate the approach’s promise. An eight-group table explains purchases and sales across industries; an eighteen-industry example traces a $435.3 million increase in demand for chemicals and munitions. Morgenstern reports $618 million in additional effects elsewhere, exceeding the initiating order. Discovering these repercussions requires solving eighteen simultaneous linear equations, involving approximately 40,000 multiplications. The article closes with qualified confidence: mathematical computation can disclose consequences unavailable to intuition, but its usefulness depends on improved data, manageable representations, and methods capable of handling simultaneous demands. Its relevance lies in treating economic programming as an interdependent decision problem while refusing to confuse calculability, feasibility, and optimality.
This work was divided into 1 sections when it entered the library's research corpus—an apparatus for search and citation, not necessarily the author's own table of contents. Each title opens its summary.
Put a question to this work; the Librarian answers from its 1 sections and cites the passage.
Ask the Librarian