Plate vii in print

The Descent

deflation; the silent trap of falling prices, the rising real burden of debt, and the floor beneath which interest cannot fall.

The promise stays written in money while the labor needed to discharge it changes.

— the argument of this plate, after Fisher and the earlier Fall

Cheaper things can be a gift. A better harvest or a more efficient workshop can put the same comforts within reach of more people, and the fall in one price need announce no general distress. Deflation poses a different question: what happens when prices fall across the economy while its promises remain written in money? In that descent, money itself gains purchasing power. The fear concerns the people who must earn it before they can pay what they owe, and the institutions trying to restore spending when households and businesses retreat together. Falling prices can accompany expanding productive capacity; they can also accompany a collapse in demand. The price index alone does not tell us which world we inhabit. The danger becomes clearer when we set the moving price beside the obligations that do not move with it.

A debt specifies a payment in a particular unit, and a change in that unit’s purchasing power changes the real claim. In a hypothetical economy, a borrower owes one thousand and a representative basket of goods costs one hundred. The principal commands ten baskets. If the basket price falls to eighty while the principal stays fixed, it commands twelve and a half: a twenty-five percent increase in the real burden after a twenty-percent fall in prices. The household’s ability to pay requires another comparison. An annual payment of two hundred takes a fifth of an income of one thousand, but a quarter if income falls to eight hundred. The first calculation measures purchasing power; the second measures cash strain. They can move together, although no accounting identity requires every borrower’s income to fall in proportion to the price level. Indexed debts, renegotiated obligations and different income paths change the result.1

1. Irving Fisher, The Debt-Deflation Theory of Great Depressions (1933), §§22–35, pp. 341–345. All numerical examples in this plate are hypothetical. Fixed principal divided by the basket price rises from 1,000/100 = 10 to 1,000/80 = 12.5; the payment-to-income ratio rises from 200/1,000 = 20% to 200/800 = 25%. Fisher also stresses maturity and the timing of payments.

Fisher’s further insight was that attempts to escape debt could worsen the conditions under which everyone had to escape it. Distressed borrowers sell assets, reduce purchases and try to repay banks; weakening sales, damaged collateral and impaired lenders can then restrict spending and credit further. The real debt can rise even as its money amount falls. Suppose the thousand of principal is reduced to nine hundred while the basket price falls from one hundred to eighty. The remaining claim is eleven and a quarter baskets, still above the original ten. That arithmetic does not establish that repayment caused the price decline; the causal argument requires the spending and financial mechanisms. Nor does it erase the creditor’s corresponding claim. The economic damage arises when strained debtors, defaults and disrupted lending reduce activity, rather than from a bookkeeping discovery that debt has no other side.2

2. Fisher (1933), complete “The Roles of Debt and Deflation,” §§22–35, pp. 341–345. His sequence is an interacting network, not an invariant chronology; §§29–33 distinguish cases in which countervailing forces or smaller indebtedness prevent the self-aggravating result. In the authored example, 900/80 = 11.25, a 12.5% real increase despite 10% nominal repayment. Neither that magnitude nor an automatic spiral is attributed to every episode.
⁂

The burden on new borrowing depends on the price change expected over the loan’s life. A zero nominal rate promises to return the same number of units, not the same quantity of goods. If a two-percent price decline over one year is anticipated with certainty, repaying one hundred when a basket has fallen from one hundred to ninety-eight costs about 2.04 percent more in real terms. Even a nominal rate of minus half a percent leaves a positive real cost of about 1.53 percent under that expectation. The familiar subtraction of expected inflation from nominal interest approximates this relationship. Unexpected deflation instead changes the realized return after the contract was made. Keeping the two comparisons separate matters: yesterday’s surprise increases the weight of existing promises, while tomorrow’s expected decline can make a seemingly cheap new loan expensive before anyone signs it.3

3. Ben Bernanke, Deflation: Making Sure “It” Doesn’t Happen Here (November 21, 2002), “Deflation: Its Causes and Effects” and notes 2–5. The exact deterministic one-period calculation is 1 + r = (1 + i)/(1 + π): 1/0.98 − 1 ≈ 2.04%; 0.995/0.98 − 1 ≈ 1.53%. The example assumes the stated price path, without default, fees or taxes; uncertain inflation requires more than treating a forecast as a guaranteed return.

The obstacle to cutting the nominal rate indefinitely is the alternative of holding currency. A banknote pays no interest, but it retains its face amount while it sits in a drawer. If keeping it were costless and lending equally safe, a lender would resist an offer to receive fewer notes later. Holding large amounts of cash is not costless, however: storage, transport and security can make a mildly negative deposit rate preferable. With an invented annual cash-carrying cost of one on a holding of one hundred, cash leaves ninety-nine after costs; an otherwise comparable deposit returning ninety-nine and a half leaves more. The effective lower bound therefore need not sit exactly at zero or at one universal negative rate. Its location depends on the available alternatives and their costs. It constrains a nominal instrument, while the real rate also depends on expected inflation.4

4. Benoît Cœuré, How Binding Is the Zero Lower Bound? (speech, May 18, 2015), introduction and “The effective lower bound is not at zero.” The cash/deposit comparison holds safety, access and other services equal and stipulates a one-unit carrying cost. It illustrates the mechanism, not an estimated bound; the speech’s dated policy settings are not presented as current.

A floor under that instrument leaves other channels open, though it makes restoring demand more difficult than turning the usual dial. Commitments about future policy can influence expected short rates; purchases of longer-term assets can affect the yields relevant to investment; fiscal action can support spending directly. Each requires an account of credibility, transmission and institutional authority. The distinction between nominal and real rates also shows why expected price stability matters: at a zero nominal rate, a certain two-percent price increase gives a real return of about minus 1.96 percent, whereas the expected decline gave a positive return. Announcing the preferred expectation does not make people adopt it. Repairing financial channels and establishing a believable policy path are part of the task. The lower bound limits one means of response; it does not establish that every remaining response is powerless.5

5. Bernanke (2002), complete “Curing Deflation,” “Fiscal Policy” and conclusion, including implementation and calibration qualifications; Cœuré (2015), “The lower bound and non-standard measures” and the later discussion of impaired transmission. The illustrative real return is 1/1.02 − 1 ≈ −1.96%. These are descriptions of channels and historical policy arguments, not assertions of present legal powers or guaranteed effects.

The descent is feared because the economy can be asked to adjust in several incompatible directions at once. Prices fall, borrowers need more purchasing power to discharge fixed promises, lenders become more cautious, and the real cost of new finance can rise while its nominal price approaches the floor. That combination can turn an effort to become solvent into a contraction that makes solvency harder. Its severity depends on the debts, the source of the price decline, the resilience of institutions and the response made in time. The argument does not rank every deflation above every inflation in harm, or require resistance to every improvement that makes goods cheaper. It asks whether a society’s commitments and its means of adjustment can accommodate the changing value of money without forcing the adjustment through lost incomes, failed enterprises and idle hands.