Inflation Does More Than Raise Prices: It Changes How Firms Plan
This series studies price-setting firms at the point where inflation becomes a decision. Its specific question is how firms respond when costs, expectations, and the economic outlook move together: do they simply pass a surprise into prices, or does the surprise also change how far ahead they are willing to think? The answer matters because firms’ choices can alter not only their own prices but also the sensitivity and uncertainty of inflation. The evidence supports a useful, bounded conclusion: inflation is partly shaped by the planning distance and information environment of the firms producing it.
That claim does not mean that firms consciously follow one universal inflation playbook. It means that the value of looking ahead can change with the state of the economy. A large or persistent disturbance makes future consequences more important. A firm with little liquidity may absorb a cost increase to protect demand, while a firm with market power may pass it through. A firm receiving reliable information about current inflation may rely less on its own cost experience when forming an aggregate forecast. These are different decisions, but they meet at the same point: a firm must decide what the present shock means for the future.
Start with the firm, not the price index
Inflation is usually presented as a movement in an aggregate price index. That measure is indispensable, but it hides the decisions that transmit a shock through the economy. A price-setting firm faces an input-cost change, decides whether its own price should change, forms a view about future costs and demand, and may use that experience to revise its beliefs about inflation more broadly. The aggregate outcome is built from many such judgments, made under different constraints and with different information.
The distinction between four objects is essential. An input-price surprise is a difference between what a firm expected its input prices to do and what actually happened. An expected own-price change is the firm’s forecast for the price it will charge. A realized own-price change is what that firm eventually does. An aggregate inflation expectation is the firm’s forecast for consumer-price inflation in the wider economy. They are related, but they are not interchangeable. A firm can expect its own costs to rise without passing the full increase into its price; it can raise its price without concluding that long-run inflation has permanently changed; and it can use a local cost experience as one signal about the economy without treating it as the whole story.
Timing matters in this chain. If a cost increase was fully anticipated, the firm may already have incorporated it into its price plan. The surprise is the part that tests whether the plan was adequate and whether it must be revised. That is why the Italian study uses the gap between realized and previously expected input-price growth rather than simply treating every observed cost increase as an unexpected shock. Even then, the resulting measure is a proxy, not a laboratory-controlled intervention. Its value comes from the survey’s repeated firm observations and from checks against demand indicators, firm characteristics, oil prices, and the euro–dollar exchange rate—not from pretending that the proxy has no assumptions.
The Italian evidence makes the first tension concrete. In the Bank of Italy survey studied by De Fiore, Lombardi, and Mangiante, the input-price forecast error is used as a proxy for an unexpected cost shock. On the paper’s standardized scale, a one-point surprise is associated with about a 0.299-point increase in expected own-price growth and a 0.187-point increase in realized own-price growth over the following year. That is incomplete pass-through: firms transmit part of the surprise and absorb, delay, or otherwise accommodate the rest. It is also asymmetric. Positive surprises produce much larger own-price responses than negative ones, while short-term aggregate inflation expectations rise mainly after cost increases.
Those estimates do not say that every firm behaves this way, or that the input surprise mechanically causes every subsequent price movement. The sample covers Italian firms with at least 50 employees, and the interpretation depends on the assumption that forecast errors capture unexpected input-cost variation rather than an unobserved demand shock. The paper reports checks consistent with that assumption. The important point for the present question is narrower: firms’ responses are a transmission mechanism, not a footnote to the aggregate data.
Two ways to see the same transmission problem
The Federal Reserve paper by Gust, Herbst, and López-Salido approaches the problem from the decision that is usually left fixed. When a firm gets an opportunity to reset its price, it recognizes that the price may remain in effect for several periods. It can reason explicitly through future contingencies for a while, but thinking through more states and consequences requires cognitive effort. Beyond the chosen window, it relies on a continuation value shaped by past experience rather than a fully state-contingent, infinite-horizon calculation.
Here, a planning horizon means the number of future periods whose contingencies the firm evaluates using the model’s structural relationships. It is not the same as a twelve-month survey forecast, and it is not simply the length of time a price contract survives. A firm can have a price that remains in effect for several periods while explicitly planning through only some of them. The horizon is a choice about analytical effort: how much of the future is worth calculating before a simpler belief becomes the better use of scarce attention.
A short horizon is not the same as ignoring the future. It means that the firm compresses the more distant future into a continuation value. In the model, that value is informed by what the firm has learned from earlier pricing problems and is updated over time. The arrangement creates a useful middle ground between a fully informed infinite-horizon plan and a purely backward-looking rule. It also explains why the same firm need not use the same planning distance in every state: the cost of thinking is comparatively fixed, but the benefit of resolving an unusually important future contingency is not.
The Federal Reserve model makes the choice sequential. The firm begins with a short horizon, asks whether incorporating one more future period would improve its price enough to justify the incremental planning cost, and continues only while the answer is yes. The benefit is larger when the omitted future consequences matter more. In the model’s analytical cases, large and persistent disturbances raise that benefit. This is the first link between inflation and planning distance: a shock changes not only the firm’s desired price, but also the value of understanding what comes next.
The two papers therefore illuminate the same firm-centered transmission problem from different sides. The Federal Reserve paper specifies a mechanism for why the amount of forward-looking effort can move. The BIS paper observes how firms’ responses vary with the direction of a cost surprise, their position in the supply chain, uncertainty, financial capacity, and the information they receive. The former is a calibrated partial-equilibrium model, not a direct observation of managerial thought. The latter is a firm-level empirical study, not a structural estimate of the model’s cognitive-cost parameter. Treating them as complementary avoids two opposite mistakes: assuming that a model mechanism has been directly measured, or treating heterogeneous survey coefficients as if they were already a complete theory.
The map is deliberately broader than either paper’s headline result. It shows where the two perspectives meet: the shock is interpreted by a firm; the firm’s planning and information choices shape prices and expectations; realized inflation feeds back into beliefs. It also leaves room for the paths to differ. A longer planning horizon can make inflation more sensitive to a shock in the model, while a financially constrained firm in the survey may absorb more of an input increase. “Firms” are not a single undifferentiated actor.
The empirical evidence also shows why a response at one horizon should not be mistaken for a permanent change in beliefs. In the Italian survey, input-cost surprises move six- and twelve-month inflation expectations, while the reported thirty-six-to-sixty-month expectations remain statistically unaffected. The Federal Reserve model describes a related distinction in different language: current shocks can receive more weight inside the planning window while a learned trend belief continues to summarize the longer run. The two results are not identical, but both warn against reading a short-term price revision as proof that long-run inflation expectations have become unanchored.
There is a second distinction in the word “uncertainty.” The Federal Reserve exercise compares a density-based, one-quarter-ahead predictive standard deviation with the size of forecast revisions; it is not measuring how much firms disagree with one another. A model can therefore match the positive co-movement between surprises and uncertainty while still explaining only a fraction of the total movement in measured risk. Keeping those meanings separate is part of understanding the mechanism, not a technical footnote.
The question that remains: how much future is worth buying?
This perspective changes how inflation uncertainty should be understood. If firms always used the same planning horizon, their response to a given shock would be governed by fixed sensitivities in the model. If the horizon changes with the state, a large disturbance can both revise expectations and induce firms to look farther ahead. The resulting longer horizon raises the model’s sensitivity of inflation to demand and supply shocks, so next-period uncertainty can rise even when the variance of the underlying shocks is unchanged.
The quantitative results give this mechanism a disciplined, not unlimited, role. The Federal Reserve model produces a positive relationship between the size of forecast revisions and one-quarter-ahead inflation uncertainty, similar to the reported data correlation of 0.43; its baseline linear-cost version produces 0.44. It implies an average planning horizon of about four quarters and very small average planning costs relative to profits. But the model’s uncertainty volatility is only about 14 percent of the empirical measure. Endogenous planning is therefore a plausible contributor, not a complete explanation of inflation risk.
The practical judgment supported by these sources is consequently conditional. When a future shock is large, persistent, or far from firms’ established beliefs, the incentive to plan may rise. When costs move upward, pass-through may be stronger than when they fall. When firms differ in market position, uncertainty, liquidity, or access to aggregate information, the same shock need not produce the same price or expectation response. Those statements are more informative than the generic claim that inflation “feeds on itself,” but they are also more limited. They identify channels and boundaries rather than offering a universal forecast.
That restraint is especially important because the two sources operate at different levels. The Federal Reserve paper asks what follows if price-setting firms choose a horizon in a calibrated model; the BIS paper asks how surveyed Italian firms’ unexpected input-cost experiences line up with their prices and expectations. Neither source alone tells us how every firm thinks, and neither removes the role of monetary policy, demand, wages, market structure, or changing shock volatility. Together they make a narrower proposition credible: the path from a shock to inflation depends on decisions about attention, pricing, and beliefs that can vary across states and firms.
The next step is to make the finite-horizon decision concrete: what exactly does a firm evaluate inside its planning window, what does it leave to experience, and when does another period of looking ahead stop being worth the effort?