At the heart of every New Keynesian model sits a variable that does more theoretical heavy lifting than any other: expected inflation. It anchors the Phillips curve, drives the Euler equation for consumption, and determines whether central bank policy rates translate into the real interest rates that actually matter for economic decisions. Remove inflation expectations from the framework, and the entire edifice of modern monetary economics collapses into something unrecognizable.

Yet here is the uncomfortable truth that practitioners and theorists alike must confront. We cannot measure this variable with anything approaching the precision our models demand. Household surveys produce answers contaminated by salience bias and numerical cognition errors. Professional forecasters cluster around consensus in ways that may reflect herding rather than information. Market-based measures embed risk premia, liquidity distortions, and segmentation effects that are themselves the subject of contentious estimation. Each instrument captures something real, but none captures the object our models actually need.

This is not merely an empirical inconvenience. It is a foundational challenge for policy design. When central banks commit to managing inflation expectations—as every major central bank now explicitly or implicitly does—they are pledging to steer a variable they observe only through a fog of noisy, contradictory signals. Understanding the nature of that fog, and what it implies for the transmission mechanism of monetary policy, is among the most consequential questions in applied macroeconomics today.

The New Keynesian Phillips curve makes the centrality of expectations mathematically explicit. Current inflation depends not just on the output gap or marginal costs, but on what firms and workers expect inflation to be in the future. This forward-looking structure means that a shift in expectations—even absent any change in fundamentals—can generate real inflationary or disinflationary pressure. The expectation is not merely a forecast; it is a causal force.

Consider the wage-setting channel. When workers bargain for compensation, their reservation wage incorporates a view about how much purchasing power they need to protect against future price increases. If they expect 4% inflation rather than 2%, they demand commensurately higher nominal wages. Firms facing higher labor costs pass them through to prices, validating the original expectation. The prophecy fulfills itself not through irrationality, but through perfectly rational optimization under uncertainty.

This self-referential quality is what makes expectations management so powerful—and so dangerous. In Michael Woodford's foundational analysis, monetary policy operates primarily through its influence on expectations about the future path of interest rates, not through the current policy rate alone. The central bank's credibility, its communication strategy, and the perceived robustness of its reaction function all matter precisely because they shape the belief structures that feed into pricing decisions.

The policy implication is striking. A central bank that has firmly anchored expectations at its target possesses an enormous stabilization advantage. Small deviations in actual inflation are treated as transitory, dampening second-round effects and reducing the sacrifice ratio associated with any necessary tightening. Conversely, a central bank that has lost this anchor faces a far steeper tradeoff, because every supply shock risks embedding itself in the wage-price dynamic through shifted expectations.

This is why the rational expectations revolution, for all its abstraction, carried such profound policy consequences. It revealed that the equilibrium a central bank achieves depends fundamentally on which belief structure the private sector coordinates upon. Multiple equilibria become possible when expectations are unanchored, and the selection among them depends on institutional credibility—a variable even harder to quantify than expectations themselves.

Takeaway

In the New Keynesian framework, inflation expectations are not passive forecasts but active inputs to wage and price decisions. Policy effectiveness depends less on what the central bank does today than on what the private sector believes it will do tomorrow.

If expectations are so central, how do we observe them? The profession relies on three broad categories of measurement, and their frequent disagreement is itself informative. Household surveys, such as the University of Michigan Survey of Consumers or the ECB's Consumer Expectations Survey, ask ordinary people what they think inflation will be. The responses are noisy, often rounded to salient numbers like 5% or 10%, and systematically biased upward relative to realized inflation. They also correlate heavily with the prices of frequently purchased items—gasoline, groceries—rather than with the broader consumption basket.

Professional forecaster surveys, including the Survey of Professional Forecasters and Consensus Economics panels, produce far tighter distributions. But this precision may be partly illusory. Forecasters face reputational incentives that reward clustering near the median and penalize outlier predictions, even when private information supports them. The result is a consensus that may understate genuine uncertainty and sluggishly incorporate regime changes—precisely the moments when accurate expectation measurement matters most.

Market-based measures, derived from inflation-linked bonds (TIPS breakevens) or inflation swaps, offer the appeal of real-time, high-frequency observation backed by financial stakes. But they embed inflation risk premia—the compensation investors demand for bearing inflation uncertainty—that must be extracted before the pure expectation component can be isolated. Estimates of this risk premium vary substantially across models, and during periods of market stress, liquidity differentials between nominal and inflation-linked bonds introduce additional distortions.

The disagreements among these measures are not random. During the 2021-2023 inflation episode, household expectations surged dramatically while professional forecasters and market measures moved more modestly. Which signal should a policymaker weight? If household expectations drive wage bargaining at the margin, their deviation from professional forecasts may be the more policy-relevant indicator—even if they are statistically noisier and less "rational" in the econometric sense.

Recent research by Coibion, Gorodnichenko, and others has pushed toward a more heterogeneous view: different agents form expectations through different information sets and cognitive processes, and no single measure captures the aggregate expectation that enters the representative-agent Phillips curve. This insight is both liberating and deeply inconvenient. It suggests the representative-agent expectation our models require may not exist as a measurable object in the real world.

Takeaway

Each expectations measure captures a different slice of a heterogeneous reality. The policy-relevant expectation may not correspond to any single observable indicator, forcing central banks to triangulate across imperfect signals while acknowledging irreducible measurement uncertainty.

Well-anchored inflation expectations are the crown jewel of modern central banking. When long-run expectations remain firmly at target despite short-run inflation fluctuations, the central bank enjoys a self-stabilizing dynamic: temporary shocks dissipate without triggering persistent wage-price spirals. The question that matters most is what determines whether this anchoring holds—and what triggers the potentially catastrophic process of de-anchoring.

Empirical evidence suggests anchoring is asymmetric and state-dependent. Expectations appear far more sensitive to upside inflation surprises than to downside surprises, a pattern consistent with loss aversion in the behavioral literature and with the greater salience of price increases in everyday experience. Moreover, the sensitivity of long-run expectations to short-run inflation news—what researchers call the "pass-through" coefficient—appears to increase nonlinearly once inflation breaches certain thresholds. Below those thresholds, anchoring is robust. Above them, it becomes fragile.

The theoretical mechanisms behind de-anchoring are multiple and reinforcing. Rational inattention models suggest that agents pay more attention to inflation when it becomes volatile and salient, causing them to update their long-run priors more aggressively. Adaptive learning models show that a sequence of forecast errors in the same direction can cause agents to revise their perceived inflation regime, particularly if they assign positive probability to structural breaks in the central bank's reaction function.

Central bank communication plays a crucial role in this dynamic. Forward guidance, explicit inflation targets, and transparent reaction functions all serve as coordination devices that help the private sector converge on the target-consistent equilibrium. But communication is a double-edged instrument. Conditional commitments that are later revised can erode credibility more than silence would have. The Fed's characterization of post-pandemic inflation as "transitory" illustrates the risk: a label intended to anchor expectations may, if falsified by subsequent data, accelerate de-anchoring by damaging the perceived quality of the central bank's information.

What makes de-anchoring so dangerous is its potential irreversibility on policy-relevant timescales. Re-anchoring expectations after a credibility loss historically requires sustained periods of restrictive policy—the Volcker disinflation being the canonical example—with large cumulative output costs. The asymmetry between the ease of losing anchoring and the difficulty of restoring it creates an overwhelming case for preemptive vigilance, even when the probability of de-anchoring appears small.

Takeaway

Expectation anchoring behaves like a ratchet: easy to lose, costly to restore. This asymmetry means the expected cost of under-reacting to incipient de-anchoring vastly exceeds the cost of over-reacting, fundamentally shaping optimal policy design.

The New Keynesian framework places inflation expectations at the center of monetary transmission, yet our ability to observe this pivotal variable remains stubbornly imprecise. Household surveys, professional forecasts, and market-derived measures each illuminate a facet of the underlying reality while leaving deep shadows elsewhere. This is not a problem that better data alone will solve—it reflects genuine heterogeneity in how economic agents form and revise beliefs.

For policy practitioners, the implication is a mandate for epistemic humility combined with strategic conservatism. Central banks must manage a variable they cannot directly see, using instruments whose effects depend on beliefs they cannot fully control. Robust policy design must account for this uncertainty explicitly, not treat it as a minor calibration issue.

The most important insight may be structural: expectations anchoring is not a permanent achievement but an ongoing equilibrium that requires continuous maintenance through credible commitment, transparent communication, and—when necessary—decisive action. The fog around expectations measurement is not going away. Learning to navigate within it is the central challenge of modern monetary policy.