The Target

A central bank acts today upon an inflation it will meet later. The prices now being recorded carry the residue of earlier decisions, contracts already signed, goods already ordered, and disturbances that policy cannot recall. An interest-rate decision enters this procession with a delay. It changes financing conditions, expenditure, and eventually the pressure upon prices; by the time its full effect arrives, the number that prompted it belongs to the past. The target therefore presents a problem of aim. A bank that pursues the latest reading alone can chase a movement that is already ending, or allow a disturbance still gathering force to pass unnoticed. The relevant question is what inflation is likely to become under the policy the bank chooses now. In this question the forecast changes its office. It becomes part of the means by which the future is governed, a conditional account of where a decision leads and a public object against which that decision may be judged.

Lars Svensson gave this arrangement its explicit form: inflation targeting can be implemented by targeting the inflation forecast.¹ The monetary authority begins with an announced inflation objective and a view of how its instrument affects the economy. It asks what inflation will be, over the horizon its action can influence, if it sets the instrument at one level rather than another. Under a strict inflation objective, it chooses the setting that brings the relevant forecast to the target. A forecast above the target calls for a different setting; a forecast below it calls for adjustment in the other direction, given the model's ordinary transmission mechanism. The procedure is forward-looking because the instrument works forward. Its apparent simplicity rests upon an important qualification: the forecast must be conditional upon the policy being evaluated. A projection made on the assumption that rates never change cannot by itself tell a bank what will happen if it changes them, and a forecast that already assumes the bank will correct every deviation cannot explain how the correction is to be made. Policy must appear inside its own account of the future.

The forecast supplies an intermediate target because the final target cannot be delivered on command.² A central bank cannot set the price of every service, stop a harvest from failing, or prevent an unexpected disturbance from arriving between decision and outcome. It can choose a policy whose expected consequence is consistent with its objective, while acknowledging that the realized consequence will also contain news. Suppose the bank's best conditional forecast returns inflation to its target, and an unforeseen supply disruption subsequently raises prices. The miss is real; it does not establish, by itself, that the earlier decision was wrong. Conversely, a favorable shock can bring inflation to target after a poorly judged decision. Assessment must distinguish the quality of the choice on the information then available from the luck attending its result. This distinction gives the forecast a second office. Published with its assumptions, it preserves the reasoning that existed at the moment of action, allowing later readers to ask which part of the error came from policy, which from the model, and which from events the decision could not have known.

A bank that also cares about employment and output must choose a path back to the target, and the speed of that return becomes part of the policy.³ A disturbance can raise inflation while depressing activity. Forcing the inflation forecast back immediately at the first controllable horizon may then require an unnecessarily severe further contraction. In Svensson's flexible formulation, the bank weighs deviations of inflation from its target together with deviations of output from its sustainable level. Greater weight upon output stabilization generally means a more gradual expected return of inflation toward the objective. The destination remains fixed while the route responds to the cost of reaching it. This is a disciplined reason for temporary departures, and it requires the departure to have a reason that can be stated. A vague assurance that the target will eventually be met supplies no such discipline. The forecast path must show the proposed adjustment, the slack it implies, and the judgment about how much disruption is warranted. The horizon is where a commitment to stable prices meets a concern for the lives through which stability is restored.

The distinction between a target rule and an instrument rule now becomes consequential.⁴ A simple instrument rule specifies how the interest rate should respond to a small set of observed quantities. A target rule specifies the condition that the projected outcome should satisfy. Both can constrain discretion, but they place the constraint at different points in the chain. When the relation between interest rates and spending changes, a mechanical response to the same current inflation may produce a different future. A forecast-targeting bank must reconsider the response while continuing to explain its decision by the same objective. New information about transmission, demand, or supply can alter the instrument setting without altering what the institution is trying to achieve. This flexibility is a strength only if the forecast and the judgment entering it are exposed to examination. An objective can remain beautifully constant on paper while a pliable private forecast excuses any action the bank happens to prefer. The accountability of the method therefore resides in the connection between its stable end and its revisable means, and in whether outsiders can inspect that connection.

Transparency gives the target its institutional force by making a conditional promise contestable.⁵ The public need not reproduce every calculation to ask whether the institution's own forecast supports its announced choice. What inflation does the bank expect? Under what interest-rate setting or path? How does the projected return differ from the alternatives, and what output cost explains that difference? A forecast without its policy assumption conceals the lever upon which its conclusion depends. A forecast without a discussion of uncertainty gives a point estimate the appearance of knowledge it cannot possess. And a forecast repeatedly revised to reconcile every decision with the same comforting destination can become a ceremony of justification. The remedy lies in an intelligible record: the assumptions, the conditional projections, the sources of uncertainty, and the reasons for choosing among feasible paths. Public criticism then has a defined object. It can challenge the estimate of the economy's condition or the weight placed upon its costs, instead of inferring the institution's purpose from the interest rate alone.

The target completes the promise made by The Anchor by giving credibility an everyday practice. An institution becomes answerable for a reasoned route from its instrument to its end, including the temporary departures that a more gradual return may require. This arrangement preserves a place for judgment, because neither the economy nor the transmission of policy holds still. It also gives judgment a burden: each departure must remain intelligible in terms of the objective, the evidence, and the costs borne along the way. The target cannot decide how heavily those costs should count. The relative weight placed upon inflation and unemployment is a social choice, and the model takes that choice as an instruction rather than producing it as a discovery. Nor can a numerical destination remove uncertainty from the journey. Its achievement is more exact and more modest: it makes a future aim present in today's decision, and gives the public a way to ask whether the decision is directed toward it. Monetary authority acquires its most durable freedom when its reasons can be followed. The argument remains open.