The CPI–Payroll Split: When Policy Uncertainty Means Different Things

This chapter uses the contrasting behavior of inflation and nonfarm-payroll news to test whether Treasury sensitivity depends on a release's diagnostic value for monetary policy. The question is why the same high-uncertainty state amplified CPI surprises after the COVID-era inflation surge but no longer had the same clear relationship with payroll surprises. The comparison gives the reader a concrete way to distinguish a change in release-specific information from a vague claim that the whole bond market simply became more volatile.

The contrast matters because CPI and payrolls are both closely watched macroeconomic releases, and both can affect expectations for interest rates. Yet “important” is not a fixed property of a data series. A release can become more or less useful for identifying the policy path as the central bank's priorities, the quality of the data, and the surrounding macro regime change.

From one pre-COVID pattern to two post-COVID responses

The BIS study first establishes a pre-2020 reference point, covering 1998 through February 2020. For most of the six releases in the sample, greater forecast disagreement dampened the Treasury-futures response to a surprise, while greater short-rate uncertainty amplified it. That pattern is already more precise than saying that uncertainty made yields move more. The relevant uncertainty was uncertainty about the future short-rate path, and the amplification applied to the release's policy-relevant surprise.

CPI was the important exception. Before the post-COVID inflation surge, CPI surprises did not become significantly more powerful when short-rate uncertainty was high across the studied maturities. The paper's interpretation is that inflation news then carried limited information about the future policy path, even when investors were uncertain about rates. The absence of amplification is not evidence that CPI did not matter at all. It is evidence that the conditional slope with respect to short-rate uncertainty was weak in that period.

Nonfarm payrolls showed the opposite pre-2020 pattern. Its response was positively conditional on short-rate uncertainty at the two-, five-, and ten-year horizons. When the market was less certain about the path of policy, a payroll surprise was more useful for inferring that path. Payrolls therefore provide a clean benchmark for the mechanism developed in the previous chapter: a release that helped identify the policy regime produced a stronger state-dependent response.

The post-COVID sample begins in July 2020, after the study excludes the most disrupted months of March through June 2020. The CPI relationship changes direction. The CPI-by-short-rate-uncertainty interaction becomes positive and highly significant across the two-, five-, and ten-year futures. Inflation news is now more powerful precisely when the policy path is uncertain. The NFP relationship weakens instead. Its previously positive interaction shrinks by roughly one-half to two-thirds and is no longer statistically distinguishable from an unconditional response.

The break is not just a visual contrast between two charts. The study's structural-break tests reject stability of the relevant CPI coefficient blocks at conventional levels, with p-values below 0.001 for the two-, five-, and ten-year CPI samples. The analogous NFP tests also reject stability, with the shift running in the opposite direction.