Expectations and the Neutrality of Money

The model is a two-period overlapping-generations economy (following Samuelson) with N young and N old traders each period; the young produce output and can sell it to the old for fiat money, but each period’s young are randomly split between two physically separated markets that cannot communicate, with the split governed by a random variable θ. This separation-with-randomization device serves two purposes: it creates a setting where information is imperfect in a precise, analyzable way, and it generates a genuine source of relative-price variation (a “real” shock) that traders must distinguish from monetary shocks — the paper’s central concern.

The nominal money supply available to the old generation evolves by m’ = mx, where x is an independent random multiplicative shock (the monetary disturbance); traders in a given market observe only that market’s equilibrium price, which is shown to depend on the state only through the ratio z = x/θ of the monetary shock to the real (market-allocation) shock, meaning the price reveals the ratio but not the two shocks separately. This is the crux of the paper’s information structure: a trader observing a high price cannot tell, with certainty, whether it reflects a high monetary shock (x), a low real-demand shock (θ), or some combination — the two are confounded whenever the underlying probability densities of x and θ satisfy a monotonicity restriction (condition 5.7) that the paper imposes to keep the confusion well-behaved.

When the real shock θ is fixed at 1 with probability one (no real disturbance), the current price fully reveals the monetary shock x to traders, and the unique equilibrium price is exactly proportional to the money supply (p = mx/y) — real quantities (output, employment, consumption, and real cash balances) are completely unaffected even by unanticipated monetary shocks, the classical neutrality-of-money result.* This establishes that imperfect information about money alone, without a confounding real shock, is not sufficient to generate non-neutrality in this framework — it is specifically the inability to distinguish monetary from real shocks that matters.

When the money supply is held constant (x = 1 with certainty) but the real allocation shock θ varies, the current price fully reveals θ to traders, and the equilibrium price and real quantities respond directly to the real disturbance, with an equilibrium relative-price elasticity between zero and one; this case is used to establish the properties later shown to hold, in modified form, in the general case with both shocks present.

In the general case where both monetary (x) and real (θ) shocks fluctuate and traders cannot separate their effects on price, Theorem 4 establishes existence and uniqueness of an equilibrium price function p = mψ(x/θ) with an elasticity between zero and one, under which an unanticipated monetary shock has genuine real effects on output and employment — money is non-neutral in the short run precisely because it is confounded with the real shock in the price signal. The Corollary immediately following shows that a fully anticipated, proportional (scale) change in monetary policy — multiplying the money-transfer variable by a known constant — leaves the equilibrium allocation completely unchanged, because traders’ ability to discriminate between real and monetary shocks (which depends on the shape of the underlying probability distributions, not their scale) is unaffected by a scale change; only genuinely uncertain, unanticipated monetary variation generates real effects.

Both generations benefit, on average, from a monetary expansion as it occurs — the old receive the transfer and enjoy higher real consumption, and the young perceive it only as a higher-than-average selling price for their own output, which on average raises their real wealth — but Lucas shows this creates an asymmetric, systematically misleading narrative: contractions that inevitably follow expansions (to keep real balances anchored) will be blamed on “the current inflation” rather than correctly attributed to the earlier expansion, even though the model’s shocks are symmetric between ups and downs. He explicitly notes this account is meant to be suggestive of a mechanism, not a substitute for econometric evidence.

Lucas fits the regression hypothesis ln(Y_t) = β₀ + β₁[ln(P_t) − ln(P_{t−1})] + ε_t (output on inflation) to realizations generated by the model economy and shows analytically that the probability limit of the estimated β₁ is strictly positive — even though, by construction, there is “no usable trade-off between inflation and real output” anywhere in the model — because the regression cannot separate the model’s genuine confusion-driven co-movement from an exploitable policy trade-off. He states that “the econometric evidence for the existence of such trade-offs is much more convincing here than is the comparable evidence from the real world,” underscoring that a strong, statistically well-behaved Phillips-curve correlation is fully consistent with there being no policy-exploitable trade-off at all.

Following Milton Friedman’s terminology, a “k-percent rule” is a monetary policy that fixes the percentage growth rate of the money supply at a constant k, with zero randomness around that path; Lucas proves that the competitive equilibrium allocation arising under any k-percent rule is Pareto-optimal (Theorem 5), meaning a deterministic, fully-anticipated monetary growth rule imposes no efficiency loss relative to what a planner with the same market/information structure could achieve. He is explicit that this result compares a k-percent rule only to other deterministic (nonrandomized) policies within the model’s market/information structure, not to policies that could somehow eliminate the underlying informational separation between the two markets — if the two markets could be costlessly merged, a superior (nonmarket) allocation would become possible, but that is outside the scope of the comparison.

Lucas explicitly frames the paper as resolving the paradox in Gurley’s parody of Friedman’s monetary theory — “Money is a veil, but when the veil flutters, real output sputters” — by constructing rational, market-clearing agents for whom a fully announced, proportional monetary expansion is neutral (money genuinely is a veil in that case), while placing them in an information environment where they cannot always distinguish real from monetary shocks, so that monetary fluctuations do move real output in the same direction (the veil “flutters”). He notes the model achieves this simplicity by abstracting from most realistic features of the business cycle, with the deliberately retained exception of the endogenous Phillips-curve-like relationship, which the paper treats as the central object to be explained rather than an unexplained empirical regularity.