THEORY-152: When a firm, public service or other organization deteriorates, its members' exit and voice are both recuperation mechanisms: voice carries richer information than exit and can be felt as a benefit rather than a cost, exit can be costly even without loyalty and can signal nothing, loyalty makes members voice rather than exit, and an organization needs floor levels of both for feedback but disintegrates or is disrupted beyond ceiling levels, while its managers may favour whichever response least threatens their hold on power

Source

Hirschman (1974), reprinted 1980, LIT-750, read in NOTE-566. The 1970 book it revisits, LIT-748, is unread.

What was actually shown

Hirschman (LIT-750) argues it from examples.

• Both are recuperation mechanisms. Voice “can and should complement and occasionally supersede exit” when organizations deteriorate.
• Voice is undervalued. It can be “a highly desired end in itself”; it carries information “rich and detailed as compared to the bareness and blankness of silent exit”; and it suits goods whose nature is uncertain to both sides.
• Exit is overvalued. It can be costly where buyers have invested in a supplier, and when dissatisfied customers churn among equally bad suppliers each gains as fast as it loses, so exit signals nothing.
• Loyalty. The essay presupposes the book’s account. What it says is that loyalty raises the cost of exit, and that where loyalty is stronger, as Hirschman reasons for two-party systems, dissatisfied members “will ordinarily voice rather than exit”.
• Floors and ceilings, and who sets them. An organization needs minimum levels of exit and voice for feedback, and cannot exist beyond maximum levels. Managers may encourage the response least dangerous to their power rather than the one that gives the most feedback.

Nothing here is tested. The examples (shareholders, parties, cities, the National Health Service, states) could have run against the claims, and Hirschman notes some that do: in Japan, permission to exit is no brake on voice.

What this does not say

• Not that voice always beats exit. Ceiling levels of voice mean disruption; the claim is that both are needed.
• Not a theory of loyalty. The essay does not redefine or develop it.
• Not a uniform sequence. Whether exit is followed by more exit or by voice “cannot be given a uniform answer”.
• Not about platforms. The churning-customers case fits multi-provider markets; applying it to platforms is the record’s reading.

Connections

• The CAIB account (THEORY-140). NASA’s engineers did not exit and did not voice; the Board describes silence. In Hirschman’s terms the organization had held voice below the floor it needed for feedback, and the Board’s remedy, an independent safety organization, institutionalizes voice. The Board does not use Hirschman’s terms.
• Defensive routines (THEORY-145). Argyris describes how voice fails to be taken up even when it is used: criticism is made untestable and its rejection blamed on the critic.
• The owner’s essay. For §6.3–6.4 it gives the members’ side of decline and managers’ interest in steering it; for §7.2, voice is acting through one’s own organization; for §8.4, floors of exit and voice are a condition of an organization learning about itself.