Plate vi in print

The Spirits

animal spirits; the volatility of investment under irreducible uncertainty, and what governs it when the future cannot be known.

A future must be acted upon before it can be known; sometimes the action helps to make it.

— the argument of this plate, after the earlier Spirits and Keynes

Investment commits the present to a future that has yet to disclose its terms. A factory must be ordered before its customers arrive, its workers trained before their product is sold, and its machinery paid for before anyone knows what a rival will invent. Some uncertainties can be narrowed by research, contracts, or insurance. Others concern changes for which experience supplies no dependable frequency. The decision still comes due. A firm may proceed, wait, shrink the project, or abandon it, and each choice creates work or withholds it from someone else. The cycle therefore depends partly on how people act where calculation remains incomplete. Confidence matters because a forecast must become an order before it can employ a worker.

Keynes placed confidence inside the investment decision, alongside the expected return itself. In the twelfth chapter of the General Theory, the state of long-term expectation includes both a forecast and the strength of the grounds for trusting it. Conventions help sustain action despite limited knowledge; animal spirits supply an impulse to undertake what a complete numerical proof cannot authorize. Yet Keynes expressly resisted making everything depend on irrational psychology, and retained a role for reasonable calculation and other influences. The distinction is useful at the level of an ordinary project. Two managers can write down the same central forecast while differing profoundly in how much weight they place on it, how much loss they can bear, and how readily they will commit.1

1. John Maynard Keynes, The General Theory of Employment, Interest and Money (1936), ch. XII, especially §§II, VII–VIII. Confidence supplements calculation; the chapter also addresses credit, conventions, and conditions that mitigate uncertainty.

A valuation makes the importance of its assumptions visible without making those assumptions true. Consider an invented project costing one hundred now and paying either one hundred fifty or sixty a year later, with no other receipts or costs. At a ten percent discount rate and an assumed sixty percent probability of the higher receipt, its expected present value is about one hundred four. Reduce that probability to forty percent and the value falls to about eighty-seven. The project crosses from a positive to a negative expected net value without any change in its physical design. These are conditional calculations. They demonstrate sensitivity to a stated belief; they supply no reason to think the probabilities known, and no evidence that a particular manager has revised them irrationally.2

2. Hypothetical expected present values: (0.6 × 150 + 0.4 × 60) ÷ 1.1 = 103.64; (0.4 × 150 + 0.6 × 60) ÷ 1.1 = 87.27, rounded. Subtracting the initial cost gives 3.64 and −12.73. These assumed probabilities illustrate valuation under specified risk; they do not resolve uncertainty about the probabilities themselves.
⁂

Beliefs can also change the return by changing what other people do. Suppose two firms can either invest or wait. Each earns six if both invest, loses four if it invests alone, and earns zero by waiting. Investment is attractive when the other firm invests; waiting is attractive when it waits. Both outcomes are self-consistent. If a firm assigns probability p to the other's investment, its expected gain from investing is ten times p minus four, so it invests when that probability exceeds two fifths. The example gives the earlier Spirits a precise coordination mechanism. It also shows what must be established before invoking one: a payoff structure with sufficiently strong complementarity. Shared optimism cannot create these two equilibria in every market simply by being shared.3

3. This new hypothetical game develops the earlier Spirits. Investing yields 6p − 4(1 − p) = 10p − 4; waiting yields zero. Both-invest and both-wait are strict pure-strategy equilibria. At p = 0.4 each firm is indifferent; below or above it, the preferred action changes.

A collective retreat can transmit belief into the incomes on which those beliefs are tested. Cancel a building project and its contractors lose an order; reduce their expected income and they may defer their own purchases. Firms elsewhere then see weaker sales, giving them a concrete reason to reconsider investment. An initial fear can acquire supporting evidence through the spending it discourages. But similar observations can also follow an underlying loss of productivity, a credit restriction, or news that correctly lowers expected demand. The visible sequence alone leaves the cause unsettled. To establish a role for confidence, an account must show what expectations changed, why they changed, and how those changes affected decisions beyond the information already contained in the economy's material circumstances.4

4. The separate Spirits plan asks what governs investment when the future cannot be calculated. The transmission described here connects it to The Multiplier. It does not infer an independent confidence shock merely from falling spending or a gloomy survey.

Restoring confidence requires an arrangement that can survive examination after the encouraging speech ends. In the two-firm example, simultaneous commitments can remove each party's fear of going alone if they are credible and enforceable. An advance purchase can secure a customer; a lender's firm offer can remove a financing contingency; a staged project can limit the resources exposed before useful information arrives. Each intervention addresses a specified obstacle. Their value depends on the project and their cost, including who bears a promise that fails. A guarantee may coordinate useful activity, but it may also transfer a bad project's losses to someone else. The fact that pessimism can sustain an inferior outcome never makes every optimistic undertaking sound.5

5. The coordination remedies are conditional deductions from the invented game, not findings about a particular policy. Credible joint investment gives each firm six; a promise that leaves unilateral loss possible does not by itself remove the original uncertainty. Project quality and the incidence of any guarantee still require separate assessment.

The spirits name the difficulty of acting on a future whose evidence is incomplete and partly formed by action itself. They become an explanation when the missing knowledge, the decision rule, and the feedback through other decisions are made visible. Otherwise confidence can become a convenient name for whatever the analyst has left unexplained. The earlier plate's warning remains: a theory must say when belief can sustain different outcomes and what could distinguish that account from its rivals. Investment may falter for good reasons, bad reasons, or reasons whose quality cannot yet be known. The next question concerns the economy's response once it does falter. Why does the loss of orders persist as idle labor while wages and prices adjust so slowly?