A Practical Model for Reading Treasury Reactions to Macro News

This final chapter answers the series' opening question: why can the same macroeconomic surprise move Treasury yields differently? The answer is that a release is not priced as a number alone. It is priced as evidence about the economic state, the future policy path, and sometimes the long-run inflation anchor, with the value of that evidence changing across information states and maturities. The goal here is to turn that conclusion into a disciplined way to interpret a yield move without mistaking it for a universal forecast or an investment instruction.

The model is deliberately modest. It does not promise that an observer can infer every latent belief in real time. It does provide a better sequence of questions than “Was the report good or bad for bonds?” The sequence begins with the surprise, separates signal quality from policy uncertainty, checks which release is diagnostic in the current regime, and then asks how far along the curve the repricing travelled.

A Treasury response is a chain of conditional inferences

Start with the release relative to expectations. The relevant surprise is the difference between the actual number and the public consensus that existed before publication. A data level can be historically high and still be unsurprising; a smaller level can be a large surprise if it differs sharply from what the market expected. The first fact to establish is therefore not whether the headline sounds strong or weak, but what changed relative to the information already in prices.

Next ask what policy question the release can answer. Payrolls, inflation, GDP, retail sales, jobless claims, and durable-goods orders do not have the same relationship to the central bank's future decisions. A positive payroll surprise may be evidence of stronger demand and a tighter policy path. A higher initial-claims surprise may instead point toward easing. Even within one release category, its diagnostic value can change when the central bank's priorities or the underlying data process changes.

Then separate two pre-release states that are often called uncertainty in ordinary commentary. Forecast disagreement concerns the quality of the signal about the data. If forecasters are widely dispersed, the consensus can be a noisier summary of private information, and the measured surprise can be a less reliable proxy for the structural information that agents actually learn from. The BIS model therefore predicts an attenuation channel: greater disagreement generally lowers the yield response in the studied sample, although its full theoretical effect can be non-monotonic.

Short-rate or policy uncertainty asks a different question: how uncertain is the future path of policy rates and the rule that maps macro conditions into those rates? When the policy path is already clear, a new release may confirm rather than revise expectations. When the path is diffuse, a release that helps identify the policy regime can be more valuable. The response can therefore be amplified without uncertainty mechanically pushing yields in one direction. A tightening signal can produce a larger increase; an easing signal can produce a larger decline.

The fourth question is about release-specific diagnosticity. The CPI–NFP comparison shows why a high-uncertainty day cannot be interpreted without asking which evidence is most useful in that regime. Before the post-COVID inflation surge, NFP's response was positively conditional on short-rate uncertainty, while CPI's was not. Afterward, CPI's interaction with uncertainty became positive across two-, five-, and ten-year maturities, while NFP's earlier relationship weakened and lost statistical significance. The pattern is not well described as a common rise in bond-market volatility. One release became more informative about the policy path while the other became less informative at the margin.

Finally, locate the move along the curve and across instruments.