Expected versus realised inflation

Every fixed nominal contract — a mortgage, a bond, a pension promise, a five-year supply agreement, a wage settlement — embeds a forecast of inflation. Nobody writes the forecast down, but it is there in the number both sides agreed to. When inflation turns out different from what that number assumed, the contract silently redistributes purchasing power from one party to the other. No default occurs, no court is involved, and the transfer can be very large.

This is the practical reason to separate expected inflation from realised inflation. Expected inflation determines prices, wages and interest rates when they are set. Realised inflation determines who was right. The gap between them — the inflation surprise — is one of the largest and least discussed wealth transfers in a modern economy.

The surprise is the transfer

Take a Swiss household with a CHF 800,000 fixed-rate mortgage, ten years to maturity, at a fixed nominal rate. The principal repayable at the end is CHF 800,000 in nominal francs regardless of what happens to prices.

If inflation runs at 1 percent a year, that CHF 800,000 is worth, in today's purchasing power, 800,000 / 1.01¹⁰ = CHF 724,231. If inflation runs at 4 percent, it is worth 800,000 / 1.04¹⁰ = CHF 540,452. The difference — about CHF 184,000 — has moved from the lender to the borrower. Neither party did anything. Inflation simply came in higher than the rate was priced for.

Run it the other way and the same mechanism explains why unexpected disinflation is brutal for debtors. A borrower who locked in a high nominal rate on the assumption that high inflation would persist, and then faced falling inflation, repays in francs that are harder to earn than expected. Debt crises in the early 1980s and again in the early 1990s owed a great deal to precisely this: nominal debt contracted under one inflation regime, serviced under another.

The general rule: unexpected inflation transfers real wealth from creditors to debtors; unexpected disinflation transfers it back. Anyone holding long-dated nominal claims — bondholders, annuitants, holders of second-pillar pensions in payment, insurers with fixed liabilities — is on the losing side of an inflation surprise. Anyone with long-dated nominal liabilities — mortgaged households, leveraged corporates, and above all governments — is on the winning side.

That last point deserves emphasis, because it is why fiscal and monetary questions cannot be cleanly separated. A government with a large stock of long-dated nominal debt has a standing, if unstated, interest in inflation exceeding what its creditors expected when they bought the bonds. The lesson on financial repression later in this course takes that mechanism apart properly.

Why the ex ante real rate is the one that drives behaviour

The previous lesson defined the real rate. The distinction here sharpens it. The ex ante real rate is the nominal rate minus expected inflation: it is what a saver believes they are locking in, and what a firm compares its project return against. The ex post real rate is the nominal rate minus realised inflation: it is the outcome.

Only the ex ante rate is observable to decision-makers at the moment of decision, so only it can influence behaviour. Only the ex post rate determines who ended up better off. A period can therefore feature high nominal rates, tight-looking policy, and yet deeply negative ex post real returns — which is roughly what happened to bondholders through much of the 1970s, when inflation repeatedly outran what nominal yields had priced.

An important corollary: you cannot infer that policy was too loose from negative ex post real rates alone. If inflation surprised on the upside, the ex ante real rate could have been positive and the policy stance defensible at the time. Judging a decision by an outcome that was not knowable when the decision was made is the classic error.

How expectations are actually measured, and what each measure gets wrong

There are three families of measures, and none is clean.

Surveys of households. Consumers are asked what they expect prices to do. These measures are systematically biased upward relative to realised inflation, and they are dominated by the prices people see most often. Research by Francesco D'Acunto, Ulrike Malmendier, Michael Weber and co-authors has shown that grocery-price exposure shapes household inflation expectations far more than the overall basket does — people effectively extrapolate from the shopping trolley and the petrol pump. Household expectations are also heterogeneous by age and by lifetime experience: Malmendier and Nagel's work on learning from inflation experiences showed that the inflation an individual has personally lived through shapes their expectations decades later, with the young updating faster because recent years are a larger share of the history they have to learn from.

Surveys of professionals and firms. The Philadelphia Fed's Survey of Professional Forecasters and the ECB's Survey of Professional Forecasters are the standard examples; in Switzerland, KOF's business surveys ask firms about their own selling-price expectations and SECO's consumer survey carries a price-expectation question. Professional forecasts are better calibrated than household ones but update sluggishly — Olivier Coibion and Yuriy Gorodnichenko's work on information rigidity documents that forecasters adjust to news more slowly than full-information rationality would imply.

Market-based measures. Where inflation-linked government bonds exist, the break-even inflation rate is the difference between the nominal and real yield of matched maturities. If a ten-year nominal bond yields 3.00 percent and a ten-year inflation-linked bond of the same issuer yields 0.80 percent real, the break-even is (1.0300 / 1.0080) − 1 = 2.18 percent. That is the inflation rate at which an investor is indifferent between the two.

The break-even is not expected inflation. It is expected inflation plus an inflation risk premium (compensation for bearing inflation uncertainty on the nominal bond) minus a liquidity premium (linkers usually trade less liquidly, depressing their price and raising their real yield). Both premia move, and they move most violently in stressed markets — which is exactly when people are most tempted to read break-evens as pure expectations. Inflation swaps, particularly the five-year forward five-year-ahead rate that the ECB has long cited, embed the same contamination.

Switzerland presents a specific problem: the Confederation does not issue inflation-linked bonds, so there is no Swiss break-even to read. The practical substitutes are the SNB's own quarterly conditional inflation forecast — conditional in the technical sense that it assumes the policy rate stays unchanged over the forecast horizon, which makes it a statement about the policy path as much as about inflation — plus KOF and SECO survey measures.

Anchoring: what it means mechanically

"Anchored expectations" is used loosely. The precise meaning is testable: long-horizon inflation expectations are anchored if they do not respond to short-horizon inflation surprises. A spike in this month's headline number should move one-year expectations and leave five-to-ten-year expectations flat. When the long horizon starts drifting with the news, expectations are de-anchoring.

The mechanism behind anchoring is credibility, and the classic theoretical account is time inconsistency. Kydland and Prescott (1977) and Barro and Gordon (1983) showed that a central bank with discretion has an incentive to generate a little surprise inflation to boost output — and that once the private sector understands this incentive, it prices in higher inflation, leaving the bank with higher inflation and no output gain. The escape is commitment: rules, mandates, independence, and a track record.

Why anchoring matters so much in practice: it determines the cost of stopping an inflation. If firms and workers believe inflation will return to target, they set prices and wages accordingly, and the return happens with modest damage to output. If they do not, disinflation requires forcing it through the labour market — the sacrifice ratio rises, and the process looks like the early 1980s in the United States rather than a routine tightening cycle.

What this means when you write a contract

Every nominal contract is a bet on inflation, so the useful question when signing one is: which side of the surprise am I on, and am I being paid for it?

A twenty-five-year second-pillar pension of CHF 4,000 a month is a long nominal claim. Its purchasing power in year 25 is 4,000 / 1.01²⁵ = CHF 3,119 at 1 percent inflation, and 4,000 / 1.03²⁵ = CHF 1,910 at 3 percent. Same contract, same francs, less than two-thirds of the real value. Since Swiss second-pillar pensions in payment carry no automatic indexation, that risk sits entirely with the retiree.

The three broad responses are: index the contract explicitly where the instrument exists; shorten its duration so it reprices; or hold the offsetting exposure elsewhere in the balance sheet. Which assets have historically done that job, and which merely claim to, is the subject of a later lesson in this course.

What to take away

• The economically meaningful quantity is not inflation but the inflation surprise — the gap between what nominal contracts assumed and what actually happened — because that gap is what redistributes real wealth.
• Unexpected inflation moves purchasing power from creditors to debtors and unexpected disinflation moves it back; on a CHF 800,000 ten-year nominal debt, the difference between 1 and 4 percent realised inflation is roughly CHF 184,000 of real value.
• Break-even inflation is not expected inflation: it is expected inflation plus an inflation risk premium minus a liquidity premium, and both premia are least stable exactly when you most want to read the signal.
• Anchoring has a testable definition — long-horizon expectations that do not move with short-horizon surprises — and it is the single variable that determines how expensive a disinflation will be.
• Switzerland has no inflation-linked federal bonds and therefore no market break-even, so Swiss expectations must be read from the SNB's conditional forecast and from KOF and SECO surveys, each with known biases.

Sources

• Finn E. Kydland and Edward C. Prescott, "Rules Rather than Discretion: The Inconsistency of Optimal Plans", Journal of Political Economy (1977).
• Robert J. Barro and David B. Gordon, "A Positive Theory of Monetary Policy in a Natural Rate Model", Journal of Political Economy (1983).
• Olivier Coibion and Yuriy Gorodnichenko, "Information Rigidity and the Expectations Formation Process", American Economic Review (2015).
• Ulrike Malmendier and Stefan Nagel, "Learning from Inflation Experiences", Quarterly Journal of Economics (2016) — the experience-effects result for inflation expectations specifically. Their earlier "Depression Babies: Do Macroeconomic Experiences Affect Risk Taking?", Quarterly Journal of Economics (2011), makes the parallel argument for risk taking rather than for inflation.
• Francesco D'Acunto, Ulrike Malmendier, Juan Ospina and Michael Weber, "Exposure to Grocery Prices and Inflation Expectations", Journal of Political Economy (2021).
• Swiss National Bank, monetary policy assessments and the quarterly conditional inflation forecast.
• KOF Swiss Economic Institute (ETH Zurich) business tendency surveys; SECO consumer sentiment survey.
• Federal Reserve Bank of Philadelphia, Survey of Professional Forecasters; European Central Bank, Survey of Professional Forecasters and Consumer Expectations Survey.