A Positive Theory of Monetary Policy in a Natural Rate Model
📄 Summarized from the full manuscript · Human-reviewed for faithfulness before publication
In brief
Why do even well-meaning, rational central banks tend to run higher inflation than seems sensible, and why do they keep intervening in the business cycle even though it does not actually lower unemployment? This 1983 paper shows both patterns can emerge from pure rationality once a central bank cannot bind its own hands in advance: the public, anticipating the bank's temptation to spring inflation surprises, raises its expectations until that temptation is exactly offset, leaving unemployment where it would be anyway but inflation permanently higher. The fix the paper points to is credible commitment -- rules rather than freedom to decide case by case.
What this paper finds — and why it matters
Barro and Gordon build a positive (not merely normative) theory of monetary policy in a natural-rate model, aiming to reconcile two stylized facts – inflation and money growth that are persistently “excessive,” and a tendency toward countercyclical monetary activism – with full rationality of both the policymaker and the public. In their model the policymaker minimizes a period-by-period loss function over inflation and unemployment, subject to an expectational Phillips-curve relationship, but cannot precommit to a monetary rule in advance; because the public understands the policymaker’s incentive to spring inflationary surprises, in rational-expectations equilibrium inflation expectations rise until the policymaker’s marginal temptation to inflate is exactly offset, leaving the equilibrium unemployment rate invariant to monetary policy (as in standard natural-rate models) while equilibrium inflation settles at a positive, “excessive” rate relative to the rate that would be chosen under commitment. The same discretionary logic generates a role for activist, countercyclical monetary responses to shocks (e.g., supply shocks, output disturbances) without those responses doing the economy any good in equilibrium, since the public’s expectations adjust to neutralize them. Barro and Gordon then show that formal or informal precommitment devices – rules, reputational mechanisms, or legislated constraints on monetary discretion – can move the economy from this time-consistent-but-inferior discretionary equilibrium to one with the same real (unemployment) outcomes but lower, less variable inflation, providing a positive-theory complement to Kydland and Prescott’s (1977) “rules vs. discretion” argument.
Summary of a classic paper, AI-assisted and human-reviewed. See the linked original for the authoritative claims and full conditions.
Questions & answers
Q1. What are the two stylized facts the paper sets out to reconcile, and why is reconciling them with rational-expectations natural-rate theory difficult?
The paper starts from two empirical observations about actual monetary policy – that average inflation and monetary growth are “excessive” relative to an efficiency criterion, and that policymakers tend to pursue activist, countercyclical monetary/inflation policy – and asks whether these facts can be reconciled with a model in which the policymaker and the public both behave fully rationally and the policymaker’s own objective reflects the public’s preferences. Within standard rational-expectations natural-rate models (e.g., Sargent and Wallace, 1975), the systematic, anticipated part of monetary policy is neutral with respect to real activity; the puzzle is why a rational policymaker with society’s own preferences would choose to run persistent countercyclical policy that, by the model’s own logic, cannot systematically affect unemployment, and why such a policymaker would tolerate “excessive” average inflation rather than simply choosing price stability.
Q2. What is the key institutional feature that distinguishes the paper’s framework from earlier natural-rate models?
Unlike earlier natural-rate models in which policy is chosen once and for all over a prespecified rule, the paper’s policymaker sets policy sequentially, period by period, without the ability to precommit to future choices – so a rational-expectations equilibrium requires not just that private decision rules be optimal given expectations, but that the policymaker’s period-by-period choices be optimal given that the public, understanding this discretionary structure, has already formed its expectations accordingly. Because agents cannot reasonably expect a discretionary policymaker to forgo an ex post temptation, equilibrium expectations must equal what the policymaker will actually find it optimal to do once those expectations are formed – the hallmark of the time-consistency problem that the paper (citing Kydland and Prescott, 1977) builds on.
Q3. Why does equilibrium unemployment end up invariant to monetary policy even though the policymaker behaves “activist”-ly?
Because the policymaker optimizes anew each period subject to the prevailing (rationally formed) inflation expectation, and because the public understands and anticipates this behavior, the only self-consistent expectation is one at which the policymaker has no further incentive to create inflation surprises – which pins down the unemployment rate at its natural-rate value regardless of the chosen inflation rate, exactly as in models with precommitted rules. The “activism” shows up only in the time path and variability of inflation/monetary growth chosen to track shocks, not in any systematic real effect, since rational private agents incorporate the policymaker’s reaction function into their own expectations.
Q4. Why is the equilibrium inflation rate “excessive” relative to a rule-bound outcome, even though unemployment is unaffected either way?
At each date, taking expected inflation as given, the policymaker still finds it marginally beneficial to generate a positive inflation surprise (since the marginal cost of inflation at zero surprise is zero while the marginal benefit from a perceived reduction in unemployment is positive), so a zero- or low-inflation expectation cannot be an equilibrium; the public, anticipating this temptation, raises its expected inflation until the policymaker’s marginal temptation to inflate is exactly offset by the marginal cost of the resulting higher inflation – a positive equilibrium inflation rate with no corresponding unemployment benefit, which is strictly worse than the outcome under a credible commitment to price stability (same unemployment, lower inflation). This is the paper’s central “positive theory” result: discretionary optimization by a rational, benevolent policymaker systematically delivers excess inflation, not because of any error or bad incentive on the policymaker’s part, but as the equilibrium outcome of the time-consistency problem itself.
Q5. What determines how high the “excessive” equilibrium inflation rate is?
The paper shows equilibrium inflation/monetary growth depends on the slope of the Phillips-curve tradeoff, the relative weights the policymaker’s objective places on inflation versus unemployment costs, and the level of the natural unemployment rate itself – so differences in these structural parameters across countries or time periods would translate, in the model, into different average rates of discretionary inflation bias.
Q6. How does the paper formalize and justify “rules over discretion” as a positive-theory, not just normative, conclusion?
The paper shows that institutional arrangements that credibly precommit the policymaker to a monetary rule – for example, costlessly enforced legislative or constitutional restrictions – can replicate the natural (unemployment-invariant) real outcome of the discretionary equilibrium while delivering lower and less variable inflation, because removing the temptation to spring surprises removes the reason inflation expectations had to rise in the first place. Because this improvement is available to a rational, benevolent public and policymaker alike (setting aside the costs of actually setting up and enforcing such institutions), the value of precommitment devices is not merely a normative preference for rules but something such a society would rationally choose if it could – hence a “positive theory” argument for the value of rules, building directly on Kydland and Prescott (1977).
Q7. What does the paper say about the limits or costs of the rules-over-discretion argument?
The authors are explicit that their conclusion that precommitment strictly dominates discretion “neglects any costs of setting up or operating the different institutional environments,” including the costs of actually enforcing precommitments – so the paper’s result establishes a benchmark comparison (same unemployment, lower inflation, under costless commitment) rather than a claim that establishing credible rules is itself costless or trivial in practice.
Key terms in this paper
Definitions below follow the paper's own usage.
- Discretion
- in the paper, the policymaker's inability to precommit to a monetary-growth/inflation rule in advance, so that policy is chosen sequentially, period by period, taking the public's already-formed expectations as given rather than as something the policymaker's own rule choice can bind; contrasted throughout with a regime of (costlessly enforced) precommitment.
- Rational-expectations equilibrium under discretion
- an equilibrium consisting of a private decision rule, an expectations function, and a sequential policy rule, such that private decisions are optimal given expectations, the policymaker's sequential choices are optimal given those same expectations, and -- crucially -- the policymaker cannot find it advantageous ex post to deviate from what the public expected, since agents would not reasonably hold expectations a rational policymaker is known to want to violate.
- Natural-rate invariance under discretion
- the paper's result that, although monetary policy is chosen sequentially and tends to respond countercyclically to shocks, the equilibrium unemployment rate is exactly the same as it would be under a precommitted rule, because rational expectations adjust until the policymaker's period-by-period optimization yields no further systematic real effect -- leaving only the inflation rate, not the unemployment rate, as the variable distorted by discretion.
- Inflationary bias (excessive equilibrium inflation)
- the paper's central positive-theory finding that the discretionary equilibrium delivers a positive, excessive average inflation rate relative to what a credible commitment to price stability would deliver, driven entirely by the policymaker's period-by-period temptation to spring an inflation surprise and the public's rational anticipation of that temptation -- not by any irrationality, error, or divergence of the policymaker's objective from society's preferences.