Inflation expectations are a feedback state
Expectations can transmit a price shock into later wage and price decisions, but “anchored” is an empirical property—not a slogan and not a forecast oracle.
- problem
- Public inflation arguments often treat expectations as either pure psychology or a single survey number, obscuring how different horizons, updating rules, shocks, policy credibility, and wage or price decisions interact.
- scope
- A small adaptive-expectations model that makes feedback and persistence inspectable, paired with Federal Reserve material on short- and long-horizon inflation expectations; it is not a forecast or policy recommendation.
- environment
- A browser-executed conceptual simulation with synthetic inflation paths, explicit parameters, and no live market or household data.
Assumptions
- Agents update an expected inflation state from past observations with one visible gain parameter.
- The synthetic path isolates expectation dynamics and does not represent a measured economy.
- Short-horizon and long-horizon expectations may respond differently to the same observed shock.
Limitations
- Real economies contain heterogeneous households and firms, contracts, sector-specific prices, fiscal and monetary policy, supply constraints, and changing credibility; one adaptive rule cannot identify those mechanisms.
- The model is educational—not investment advice, a policy prescription, or a prediction of any country’s inflation path.
- Survey expectations, market inflation compensation, professional forecasts, and model-implied expectations are different measurements and should not be substituted for one another.
Table of contents 4 sections
Treat expectations as state, not commentary
An expectation is useful in a model only when its update rule is explicit. The simplest adaptive form stores a state for expected inflation and moves that state toward the latest observation. The gain parameter controls how quickly recent inflation displaces the prior belief. A low gain produces inertia; a high gain follows shocks quickly and also forgets them quickly once observations normalize.
This is not a claim that households or firms literally run one equation. It is a diagnostic instrument. It forces the argument to name what is remembered, how strongly new evidence is weighted, and which horizon is under discussion. It also exposes why one short-term survey jump cannot by itself prove that long-term expectations have de-anchored.
\pi_t^e = (1-\alpha)\pi_{t-1}^e + \alpha\pi_{t-1}, \qquad 0 \le \alpha \le 1The update gain reshapes persistence
The synthetic input holds inflation at 2, jumps to 6 for two periods, then decays toward 2. With a gain of 0.65, expected inflation reacts strongly and returns faster after the shock. With a gain of 0.20, the initial response is smaller but the expectation remains elevated for longer. The same observation path therefore produces different timing even before adding policy, contracts, or strategic price setting.
Run the program and change alpha or the path. The useful question is not which line is “correct.” It is which behavior an empirical expectation measure actually supports, over which horizon, for which population, and under what information set.
loading plot…
sandbox ready; edit the program and run it
Anchoring is a measurement claim
The Federal Reserve material emphasizes horizon. Short-term expectations react to salient prices and current shocks; longer-term expectations are expected to move less when credibility is intact. Calling expectations anchored therefore means testing their sensitivity and distribution, not observing that one median remains near a target.
The measurement surface should retain who was asked, the forecast horizon, the uncertainty or distribution, the reference price index, and whether the measure came from a survey, market price, professional forecast, or structural model. Those channels answer related but non-identical questions.
| measure | useful signal | important limitation |
|---|---|---|
| household survey | experienced prices and household plans | salience, basket differences, and wide uncertainty |
| firm survey | pricing, cost, and wage-setting intentions | sector mix and strategic reporting |
| market compensation | priced inflation plus risk and liquidity premia | not a pure expectation |
| professional forecast | model- and information-rich central outlook | small population and shared model risk |
Keep the policy boundary visible
Expectations can influence current decisions, and policy credibility can influence expectations. That does not make communication omnipotent or shocks imaginary. Energy, supply chains, labor markets, fiscal choices, exchange rates, contracts, market structure, and distributional exposure remain causal parts of the system.
The responsible use of the small model is comparative: vary an assumption, inspect the response, and state what evidence would distinguish the alternatives. It should make uncertainty easier to see, not convert a diagram into a policy conclusion.
Why this article avoids a current-market forecast
A defensible forecast needs current data vintages, a specified economy and price index, an empirical model, scenario assumptions, uncertainty intervals, and a review date. The conceptual simulation intentionally supplies none of those, so it is presented only as a feedback-model companion.
- sourced Federal Reserve discussion of expectation horizons
Governor Adriana Kugler’s 2025 speech distinguishes volatile short-horizon expectations from longer-horizon expectations and explains anchoring as relative insensitivity to incoming shocks.
inspect source ↗ - inferred Transparent adaptive-expectations simulation
The plotted paths are generated from the equation shown in the article and a synthetic shock sequence. They demonstrate model behavior, not an empirical estimate.
- reference Inflation Expectations and Monetary Policymakingopen ↗
Federal Reserve speech covering the Phillips-curve framing, horizon differences, survey dispersion, and the meaning of anchored expectations.
- demo Adaptive-expectations sandboxopen ↗
An in-page, network-disabled JavaScript runner for changing the update gain and synthetic shock path.