The Temptation — The Level, Plate v

The Phillips curve showed that inflation can buy a short burst of employment by surprise; the natural-rate hypothesis showed the burst to be fleeting and the inflation lasting. Set these two truths side by side and a disturbing problem comes into view, one that afflicts even a wise and well-meaning central bank: a structural temptation to inflate, which biases policy toward inflation even when everyone involved knows better. Finn Kydland and Edward Prescott named the problem in 1977 — the time-inconsistency of policy, the way a plan that is optimal to set today can betray itself tomorrow — and in naming it made the decisive case for taking the power over money's value out of the realm of free discretion altogether. The next step is to see why even a good central bank cannot be trusted with an unconstrained power to inflate.

Kydland and Prescott identified a general flaw in discretionary policy of any kind: a course of action that is best to promise in advance may not be best to follow through on when the moment to act arrives. Apply the principle to money. Before workers and firms set their wages and prices, the central bank's best move is to promise low inflation — for if people expect low inflation, they set modest wage demands, and the economy enjoys low inflation with employment at its natural rate, the happiest outcome available. But once the wages have been set on the strength of that promise, the bank's incentive quietly changes. Now a surprise inflation would, by the Phillips mechanism, push employment above the natural rate, since the wages are locked in too low in real terms. So the bank is tempted to renege — to spring the inflation it promised to forgo. The low-inflation policy was optimal to promise; it is not, once expectations are set, optimal to keep. The promise and the follow-through pull in opposite directions, and that is the inconsistency over time.¹

1. A flaw in discretionary policy: a course best to promise in advance may not be best to follow through on. Before wages are set, the bank's best move is to promise low inflation — for if people expect it, they set modest wage demands, and the economy gets low inflation at the natural rate. But once wages are set on that promise, a surprise inflation would push employment above the natural rate, so the bank is tempted to renege. The low-inflation policy is optimal to promise but not, once expectations are set, to keep.

The catch is that people are not fools. Knowing the bank will be tempted to renege, they do not believe its low-inflation promise in the first place. Anticipating the surprise inflation to come, they build it into their wage demands from the start, setting wages on the expectation of higher inflation. Now the bank is trapped. If it delivers only the low inflation it promised, real wages are too high — for people had expected and priced in more inflation — and unemployment rises above the natural rate. To avoid that, the bank validates the higher inflation everyone expected. So the economy settles at higher inflation — precisely the inflation the public anticipated — with employment back at the natural rate regardless. The outcome is the worst of both worlds: the higher inflation, and none of the employment gain that was supposed to justify it. This is the inflation bias of discretionary policy. A central bank free to inflate will be expected to inflate, and so will be forced to deliver an inflation from which it reaps no benefit at all. The very freedom to do good by surprise is what guarantees the worse result.²

2. But people are not fools. Knowing the bank will be tempted to renege, they disbelieve the low-inflation promise and build higher inflation into wages from the start. Now the bank is trapped: deliver only low inflation and real wages are too high, raising unemployment; so it validates the expected higher inflation. The economy settles at higher inflation with employment at the natural rate anyway — the worst of both. This is the inflation bias: a bank free to inflate is expected to, and so must, with no gain.

⁂

The remedy follows directly from the diagnosis. If the problem is that the bank is free to succumb to the temptation, the solution is to take the freedom away. Bind the central bank to a rule — a commitment to low inflation that it cannot break — or design its institutions so that it will not break the commitment even when tempted. If the bank truly cannot inflate, or credibly will not, then the temptation is moot; and people, knowing the bank cannot renege, believe the low-inflation promise at last. Expectations are anchored low, wage demands are modest, and the economy reaches the good outcome the discretionary bank could never attain: low inflation with employment at the natural rate. The paradox is exact and instructive — the bank achieves more by being able to do less. By surrendering the freedom to inflate, it gains the credibility that delivers low inflation. Rules beat discretion here not because a rule is wiser than a thoughtful official in each passing moment, but because the power to deviate is itself the source of the bias; remove the power, and the bias goes with it.³

3. The remedy follows from the diagnosis: take the freedom away. Bind the bank to a rule it cannot break, or build institutions that will not break it. If the bank credibly cannot inflate, the temptation is moot; people believe the low-inflation promise, expectations anchor low, and the economy reaches the good outcome discretion could not. The paradox: the bank achieves more by being able to do less — surrendering the freedom to inflate buys the credibility that delivers low inflation. The power to deviate is itself the source of the bias.

This is the deepest argument for an independent central bank, and it inverts the usual worry about power. One might suppose the danger of the power to inflate is that bad actors will abuse it; the inconsistency argument shows the danger to be subtler and more troubling — the power tempts even good actors, and merely possessing it, with the freedom to use it, produces the inflation bias regardless of anyone's intentions. The answer is to insulate the power from the temptation: to lodge it in a central bank independent of the government of the day, staffed by officials with long terms and a narrow mandate for stable prices, shielded from the political pressures that sharpen the temptation — the urge to inflate before an election, or to inflate away the burden of the government's debts. An independent, mandate-bound central bank is, in this light, a commitment device: a way for a society to tie its own hands, to make its low-inflation promise credible by removing from the decision the very actors most tempted to break it. The empirical record bears the argument out — across countries, the more independent the central bank, the lower the average inflation, and with no penalty in lost growth to show for the discipline.⁴

4. This is the deepest case for an independent central bank, and it inverts the usual worry: the danger is not that bad actors abuse the power but that it tempts even good ones — merely having it produces the bias. So insulate it: a bank independent of the government, long terms, a narrow mandate for stable prices, shielded from the pressure to inflate before elections or away debts. Independence is a commitment device — a society tying its own hands. The record agrees: more independence, lower inflation, no cost in growth.

The inconsistency argument brings the volume to the threshold of its central question. The power over the value of money is so structured that it cannot safely be left to discretion, even good discretion; it must be constrained, committed, insulated — held by an institution designed precisely so that it will not do the harmful thing it would otherwise be tempted to do. But this raises the deepest puzzle of all, the one the flagship confronts. If money's value rests, in the end, on the credibility of the institution that governs it — on the belief that it will keep its promise — then what anchors the value of money is not gold, nor any physical thing, but a commitment sustained by design and believed by the public. The answer to the author's question, who should hold the power to change the value of every price at once, is beginning to take shape: not the voters directly, nor a government free to choose afresh each year, but a credibly committed, insulated institution — an anchor. The next plate is the anchor.⁵

5. The argument brings the volume to its central question. The power over money's value cannot safely be left to discretion; it must be constrained, committed, insulated — held by an institution built so it will not do what it would otherwise be tempted to do. And if money's value rests on the credibility of that institution — on belief it will keep its promise — then what anchors money's value is not gold but a commitment, sustained by design and believed. The answer to who should hold the power takes shape. Next: the anchor.

Even a good central bank faces a structural temptation to inflate, and the temptation biases discretionary policy toward inflation. Kydland and Prescott showed why: low inflation is optimal to promise but, once wages are set, tempting to break by a surprise inflation — and people, anticipating the broken promise, build higher inflation into their expectations, so the bank is forced to deliver an inflation it gains nothing from. The freedom to inflate guarantees the inflation bias. The remedy is to remove the freedom — to bind the bank to a rule, or make it independent and mandate-bound, so that it cannot succumb, which makes the low-inflation promise credible and delivers low inflation at last. The power over money's value, in short, is safest when it is constrained, committed, and insulated from the very pressures that would abuse it. What this means for the nature of money's value, and for who should hold the awesome power to set it, is the volume's culminating question. The next plate is the anchor.