Oskar Morgenstern, The Compressibility of Organizations and Economic Systems (1954)

Oskar Morgenstern’s Project RAND Research Memorandum RM-1325, dated 17 August 1954, develops a conceptual vocabulary for understanding how economies and organizations can lose activities yet retain essential functions. Its four sections move from the problem’s definition through an extended analysis of basic concepts to practical applications and a tentative research program. “Compressibility” denotes selective reduction, not simply diminished output: expendable activities can disappear, or their resources can be redirected, while a system continues operating in some basic form. Morgenstern seeks to identify the conditions separating survivable reduction from collapse, without claiming to have supplied a finished theory or a reliable procedure for locating those limits.

The argument turns on the “kernel”: a configuration whose remaining parts are indispensable to the functions being preserved. Some systems have an identifiable core surrounded by dispensable layers; others contain several possible cores that emerge under different pressures. Essentiality is therefore not necessarily an intrinsic property of a component. It can depend on the configuration in which that component operates and on the substitutions still available.

A kernel begins to appear if further compression in the same direction brings about a discontinuous contraction of the performance of the remaining part of the system.

The discontinuity distinguishes a kernel from an arbitrarily reduced organization. Compression approaches a point at which further losses no longer produce manageable deterioration but destroy the relevant system-level performance. Different directions of compression may reveal different kernels. In redundant systems, resilience rests less on protecting one permanently identifiable center than on preserving relations that permit compensation and substitution. Once these alternatives have been exhausted, the remaining configuration becomes totally vulnerable.

Morgenstern uses this distinction to question economic models that impose excessively uniform interdependence. He treats the input-output coefficient matrix as totally connected, arguing that its implied vulnerability cannot adequately represent an economy’s actual resilience; at best, under suitable aggregation and technological assumptions, it might describe one kernel. By contrast, von Neumann’s expanding-economy model permits the loss of a process to reduce growth rather than immediately destroy the system, and may contain no kernel at all. These comparisons establish a methodological demand: mathematical connectedness must be tested against technological dependencies rather than mistaken for their sufficient description.

The typology distinguishes physically identifiable kernels with chain-like or star-like outer layers, redundant systems with alternative kernels, and an intermediate case in which some activities belong to every possible kernel. These arrangements imply different degrees and directions of dependence. The central conceptual move is away from universal, simultaneous interdependence toward differentiated relationships among activities.

The vulnerability of a redundant system, as described in the preceding paragraph, does thus not rest in a particular component, but in the relation of components to each other.

Compression also requires adjustment. Prices may eliminate demands that can no longer be met, while central authorities may select processes through priorities, quotas, and production programs. Neither mechanism removes the need for technological knowledge. Nor can a kernel be defined without a time horizon: deferred upkeep, repair, or adequate nutrition may permit temporary survival but eventually cause breakdown without any additional external pressure. What is viable briefly may be unsustainable indefinitely.

Four illustrations—communications, military units, organisms, and civilian administration—show how differently compression can work. Administrative shortcuts may remove formalities while requiring surviving offices to alter their behavior and relationships; organisms offer extensive compensation before reaching breaking points. Crucially, collapse is relative to the organized function under examination. A division can cease to exist as a division while smaller units continue fighting, just as an economy can lose its integrated character while regional activities survive. The question is not whether every component becomes inactive, but whether the relevant collective performance remains possible.

The discussion of artifacts, saving, and growth further separates compression from mere deprivation. Stripping a vehicle of conveniences can preserve transportation; removing equipment before battle can improve suitability for action. Active compression cuts back an existing system, whereas passive compression prevents expendable layers from expanding. Resources released in either way may support reserves or kernel growth, but the existence of alternative kernels leaves the direction of expansion undetermined.

A definition of "essential" services implies a technology.

This qualification prevents “essential” from becoming a timeless classification of goods. Requirements vary with productive techniques, organizational arrangements, activity levels, and the system’s environment. Passenger cars, for example, may be dispensable in one economy but necessary for workers to reach essential industries in another. Compression also produces compensating burdens: reduced new construction can increase repair needs. Such “bulges” may be manageable briefly yet make a prolonged reduction impossible. Accordingly, compression differs from contraction: it selectively preserves or favors functions, whereas an unguided contraction may damage potential kernels while leaving outer layers relatively intact.

The applications concern reconstruction after wartime damage, protection of a basic economy, reserve provision, and logistical systems capable of shedding nonessential services under stress. Apparently luxurious goods may support indispensable functions, making intuitive distinctions unreliable. Morgenstern’s broader challenge to economics is explicit:

It is quite possible that price, money and exchange features have been stressed far too much in economic analysis, and that technology has been quite unduly neglected.

The concluding research program proposes simple models, computational exploration, and historical study of organizations under pressure to identify breaking points. Its relevance lies in connecting organizational resilience, selective resource allocation, and technological dependence. Yet Morgenstern ends by acknowledging that the concepts remain tentative and that even the existence of some proposed problems is uncertain. The memorandum offers a research agenda for discovering minimal viable configurations, not a demonstrated method for calculating them.