Scheduled inflation releases concentrate new information into a known moment. That makes CPI announcements a useful laboratory for studying how markets process macroeconomic surprises.
This study is designed to estimate the direction, magnitude, and persistence of S&P 500 reactions following U.S. Consumer Price Index releases. It separates the announced inflation level from the surprise relative to expectations and conditions the response on the pre-release volatility regime.
The artifact presented here is a preregistration-style research protocol. It defines the question, variables, event windows, tests, and limitations before the final dataset is analyzed. Results will be added only after data validation and reproducible code review.
How do the size and direction of CPI surprises affect S&P 500 returns over one-, five-, and ten-trading-day windows—and does the pre-release volatility regime change that response?
Measures what the market learned, not simply the reported inflation rate.
Tests whether uncertainty amplifies or dampens the price response.
Tracks immediate reaction and short-horizon persistence.
NEGATIVE SURPRISE RESPONSE
Higher-than-expected inflation is associated with lower same-day S&P 500 returns, on average.
DIRECTIONAL / PRIMARYVOLATILITY AMPLIFICATION
The absolute return response is larger when pre-release implied volatility is elevated.
CONDITIONAL / PRIMARYRESPONSE DECAY
The relationship weakens across the five- and ten-day windows as other information enters prices.
HORIZON / SECONDARYASYMMETRIC REACTION
Upside inflation surprises create a different magnitude of response than equally sized downside surprises.
NONLINEAR / EXPLORATORYEVENT STUDY,
WITH THE ASSUMPTIONS
LEFT VISIBLE.
Release selection
Monthly U.S. CPI releases over an approximately 80-event history. Each event receives a timestamp, release date, actual value, expected value, and prior value.
Standardized signal
The raw surprise is actual minus consensus. A standardized version divides by the rolling dispersion of past surprises to improve comparability through time.
Response windows
Calculate close-to-close and, where the timestamp permits, open-to-close S&P 500 returns for 1D, 5D, and 10D forward windows.
Volatility condition
Classify events using the distribution of pre-release VIX levels. Sensitivity analysis will compare binary and quantile-based regime definitions.
Model family
Begin with difference-in-means and nonparametric tests, then estimate regression specifications with interaction terms and robust standard errors.
Robustness
Test alternate event windows, winsorization rules, surprise definitions, and influential observations. Report sensitivity rather than selecting one convenient specification.
Rt→t+k = α + β₁·Surpriset + β₂·HighVIXt + β₃·(Surprise × HighVIX)t + εt Where k ∈ {1, 5, 10} trading days. Coefficient β₃ tests whether volatility regime changes the relationship between the inflation surprise and subsequent return.
SOURCE GOVERNANCE / Final paper will record download dates, transformation logic, missing-value decisions, revisions, and a machine-readable data dictionary.
FROM RAW RELEASE
TO DEFENSIBLE CLAIM.
PIPELINE / 01—06INGEST
Collect releases and market series
VALIDATE
Audit dates, timestamps, and missingness
ENGINEER
Create surprises, regimes, and returns
DESCRIBE
Inspect distributions and influential events
ESTIMATE
Run primary and sensitivity specifications
REPORT
Publish results, code, and limitations
ILLUSTRATIVE OUTPUT DESIGN / Points and intervals above are placeholders showing the planned reporting format. They are not empirical estimates.
SMALL EVENT SAMPLE
Monthly releases provide relatively few observations, limiting power and the complexity of defensible models.
OVERLAPPING INFORMATION
Other macro news, earnings, geopolitical events, and policy communication can affect returns within the same window.
CHANGING REGIMES
The meaning of an inflation surprise depends on the policy and growth environment; a stable full-sample coefficient may hide structural change.
CONSENSUS QUALITY
Historical expectation data may vary by provider and snapshot timing, making provenance essential.
THE NEXT QUESTION
IS PART OF THE RESULT.
- 01Compare equity response with Treasury yields, sector ETFs, and style factors.
- 02Model the full intraday response when high-frequency data becomes available.
- 03Test whether text from CPI releases adds information beyond the headline surprise.
- 04Compare CPI reactions across monetary-policy regimes and inflation eras.
U.S. Bureau of Labor Statistics. Consumer Price Index release documentation and historical tables.
Federal Reserve Bank of St. Louis. FRED economic data series and metadata.
Cboe Global Markets. VIX methodology and historical index data documentation.
MacKinlay, A. C. Event studies in economics and finance. Journal of Economic Literature.
Campbell, Lo & MacKinlay. The Econometrics of Financial Markets.
RESEARCH PROTOCOL / v0.1
Use your browser’s print dialog to save this complete protocol as a PDF.