Why a Macro Surprise Does Not Have a Fixed Yield Impact
This series examines the U.S. Treasury yield curve's changing response to macroeconomic surprises: why the same kind of data release can move two- and ten-year yields by very different amounts, or sometimes barely move one of them at all. The opening question is not whether markets react to news. It is what a yield move says about investors' beliefs concerning future policy and inflation. By the end, a reader should be able to separate the surprise itself from the information value the market assigns to it.
That distinction matters whenever a headline says that a strong employment report or a higher-than-expected inflation reading “sent yields higher.” The description may be directionally correct, but it leaves out the condition that determines the size of the move. A release is not priced in isolation. It is filtered through the market's prior expectations, its uncertainty about the central bank's next decisions, and its judgment about whether the release is a reliable guide to the path of rates.
A surprise is a gap, not a direction
A macroeconomic surprise is the difference between an actual release and the forecast that was publicly available before it. If economists expected payrolls to rise by 150,000 and the reported increase was 200,000, the surprise is the 50,000 gap—not the 200,000 level by itself. Researchers usually standardize such gaps so that surprises in payrolls, consumer prices, retail sales, or GDP can be compared in a common statistical unit.
The sign of that gap does not determine the sign of every Treasury move. A stronger-than-expected procyclical release may lead investors to expect tighter monetary policy and push yields up. A higher-than-expected initial-jobless-claims figure can instead suggest economic weakness and a greater chance of easing, so the corresponding yield response can run in the other direction. Even this directional description is only a starting point: the market is pricing an expected path of policy, not mechanically translating “good” or “bad” news into a bond price.
The maturity of the security adds another layer. A two-year Treasury is especially sensitive to the expected path of policy over the near term. A ten-year yield reflects an average of expected short rates over a much longer horizon, together with compensation for inflation and other risks. A release can therefore have a large effect at the front of the curve while having a smaller effect farther out, or can alter long-run beliefs enough to move the long end even when the expected policy move is temporary.
The important object is thus a conditional response: how much does a particular maturity move when a release differs from consensus, given what investors already know and do not know? That wording may sound technical, but it describes a familiar market judgment. A piece of news matters more when it changes an important belief and less when it confirms an expectation or is too noisy to guide one.
Consensus also matters because the release is a public signal arriving at one moment. Investors do not first discover the number one by one and then trade around a private interpretation; they receive the same announcement and revise beliefs from a common starting point. The market's baseline is therefore part of the event. A large gap from a well-agreed forecast can be more informative than the same gap from a forecast that was already scattered, even before anyone decides whether the result is good or bad for bonds.
The long-rate puzzle makes the question consequential
The reason to study this conditional response is that long-term rates have historically reacted to news that standard macroeconomic models treated as temporary. A 2003 Federal Reserve study examined U.S. forward rates—the interest rates implied for borrowing at future dates, rather than simply the average yield to maturity. If a shock only moved short-term rates for a limited time, a forward rate far in the future should have changed little when long-run inflation and real-rate conditions were assumed to be fixed and known.
The data did not fit that simple prediction. In the study's main U.S. sample, many macroeconomic releases significantly affected one-year forward rates ending five and ten years ahead. Eleven variables were significant at the five-year horizon and ten at the ten-year horizon, and in many cases the long-horizon response was only modestly smaller than the response of the one-year rate. Unexpected monetary-policy decisions also moved long forward rates, sometimes in the opposite direction from short rates. The result was called “excess sensitivity”: long-horizon market rates appeared to contain more revision of long-run beliefs than the benchmark models allowed.
That distinction between a yield and a forward rate is important. A long yield averages the cost of borrowing across many future periods, while a forward rate isolates the rate the market implies for a particular future interval. A move in the ten-year yield can therefore reflect a collection of revisions along the path; a move in a ten-year-ahead forward rate is a sharper sign that beliefs about the distant part of that path changed. Neither measure is a pure reading of expected short rates because risk compensation and measurement choices also matter, but the distinction prevents a historical finding about forwards from being casually presented as a current ten-year-yield estimate.
The study's additional evidence points toward inflation expectations as an important part of the explanation. Long-term forward rates derived from inflation-indexed Treasury securities generally showed little response to the same surprises, while nominal-rate responses were more closely associated with inflation compensation. The authors also found that the United Kingdom's long forward rates became much less responsive to macroeconomic news after the Bank of England gained operational independence in 1997, a period when its explicit inflation target became more credible. These comparisons do not prove that every nominal-rate move is an inflation-expectations move; the TIPS sample was short, and expected inflation cannot be perfectly separated from inflation risk premia. They do show why the credibility of the long-run policy anchor belongs in the analysis.
The later BIS study makes the older puzzle more specific. Using intraday changes in two-, five-, and ten-year U.S. Treasury futures around six scheduled macroeconomic releases, it finds that the slope from a surprise to a yield response changes with the information state. Greater disagreement among forecasters generally dampens the response. Greater short-rate uncertainty generally amplifies it. Before the post-COVID inflation surge, however, CPI surprises were not amplified by short-rate uncertainty in the same way as most other releases. Afterward, CPI became positively conditional on that uncertainty, while the earlier relationship for nonfarm payrolls weakened.
These findings turn a broad historical puzzle into a sharper question: what makes a release informative about the policy path in one period but less informative in another? The answer cannot be read from the surprise's sign alone. It requires looking at the quality of the signal, the uncertainty about how policy will respond, and the horizon over which the market is repricing.
The response depends on information, not just the number
A useful way to think about the Treasury market is that it is constantly asking two different questions. First, how reliable is this release as evidence about the underlying economy? Second, how much will the central bank's future decisions change if that evidence is true? Forecast disagreement bears on the first question. If forecasters' estimates are widely dispersed before the release, the consensus may be a noisy summary of private information, and the eventual data print may be a less precise guide to the state of the economy.
Short-rate uncertainty bears on the second question. When investors have a settled view of the policy path, a new data point may add little even if it is surprising. When the path is unclear, a release that helps distinguish possible policy responses can be more valuable. This does not mean that uncertainty always pushes yields up. It means that uncertainty can increase the magnitude of the market's response to news that is relevant to resolving that uncertainty. The direction still depends on what the release implies for future rates.
This framework also explains why the same macro category can change meaning across regimes. The BIS evidence attributes the post-COVID CPI/NFP split to changes in release-specific diagnostic content: inflation became more informative about the policy path as the Federal Reserve placed greater weight on price stability, while payroll data became a noisier or less distinctive signal after labor-market disruptions and the rise of inflation as the marginal policy concern. The paper's fitted model finds a large increase in CPI's regime-mean separation and a decline in NFP's, while the estimated policy-slope spread is statistically similar across the two periods. That is a disciplined interpretation, not a claim that the model has observed investors' beliefs directly; its joint specification test is rejected and several extensions remain open.
The practical consequence is a change in how a yield move should be read. A surprise tells us what changed relative to expectations. It does not, by itself, tell us how much the market learned about future policy, whether that learning was credible, or whether the move belongs mainly to near-term rates, real rates, or long-run inflation compensation. A high-disagreement release may be discounted because it is noisy. A high-policy-uncertainty release may be amplified because it resolves an important question. A long-maturity nominal move may additionally signal changing confidence in the inflation anchor.
This is also why the evidence should be used to interpret sensitivity, not to manufacture a simple forecast. The BIS event study uses a narrow announcement window and reports how the slope of the response varies with pre-existing conditions. It does not claim that disagreement or uncertainty alone caused every observed move. The Federal Reserve evidence gives a historical explanation for why long-run beliefs can matter, but its earlier sample does not supply a permanent coefficient for today's market. Keeping those boundaries visible makes the framework more useful: it tells us which questions to ask before assigning a narrative to the curve.
The rest of this series develops those relationships without treating them as interchangeable. It first makes the event-study measurements precise, then follows the learning channels that produce opposite effects for disagreement and policy uncertainty. It tests the logic against the CPI–payroll divergence, and finally returns to long rates, real rates, and inflation anchoring. The central judgment is modest but useful: to understand why a macro surprise moved Treasury yields differently this time, ask not only what the number was, but what the market believed the number could reveal.