[DSRP Evidence](https://dsrpevidence.org/)

# Aggregation of Variables in Dynamic Systems

## Details

**Authors** Simon, Herbert A. and Ando, Albert

**Year** 1961

**Publisher** Econometrica

**Kind of work** article

**Discipline** Economics

**Secondary disciplines** Systems Science

**Applied** false

[Read it at the publisher](https://doi.org/10.2307/1909285) 
10.2307/1909285

## In authors' words

### Abstract

Previous studies have examined dynamic systems that are decomposable into independent subsystems. This article treats of systems that are nearly decomposable--systems with matrices whose elements, except within certain submatrices along the main diagonal, approach zero in the limit. Such a system can be represented as a superposition of (1) a set of independent subsystems (one for each submatrix on the diagonal) and (2) an aggregate system having one variable for each subsystem. This superposition separates short-run from long-run dynamics and justifies the ignoring of 'weak' linkages in partial equilibrium studies.

### What they set out to do (purpose)

To extend earlier work on dynamic systems that decompose into independent subsystems to systems that are only nearly decomposable.

### Who or what was studied (sample)

No empirical sample. Mathematical analysis of dynamic systems represented by matrices.

### How they did it (methods)

Formal analysis of systems whose matrices have elements approaching zero in the limit everywhere except within certain submatrices along the main diagonal.

### What they found (results)

A nearly decomposable system can be represented as a superposition of a set of independent subsystems, one for each diagonal submatrix, and an aggregate system with one variable for each subsystem. This superposition separates short-run from long-run dynamics and justifies ignoring weak linkages in partial equilibrium studies.

## Commentary

### In short

Simon and Ando showed how to simplify a large dynamic model whose parts interact strongly within clusters and only weakly between them. The model splits into the clusters' own short-run behavior and a long-run system in which each cluster acts as a single variable.

### Response

The paper gives a formal version of S and R working together. A system is decomposed into subsystems according to the pattern of its relationships, strong within blocks and weak between them, and a new whole, the aggregate system, is built with one variable per subsystem, so each part becomes a single element at the higher level. The separation of short-run from long-run dynamics is a case of behavior that depends on the level of description, as the part-whole pattern predicts. Caveats: the result is a theorem about dynamic systems and holds only to the extent that cross-block linkages are weak; it says nothing about cognition; and deciding which links count as weak is a modeling choice that could be read as a D or P question, which the paper does not frame that way, so only S and R are claimed.

**Patterns it shows** S, R

Formal proof or model

**Added** 2026-11-19

**How to cite this** Simon, Herbert A. and Ando, Albert (1961). Aggregation of Variables in Dynamic Systems. Econometrica.
