---
title: "What is an Oil Shock?"
authors: "James D. Hamilton"
source: "Journal of Econometrics"
source_url: "https://www.sciencedirect.com/science/article/abs/pii/S0304407602002075"
paper_id: "J. Econometrics 113(2), 363–398"
harvested: 2026-09-01
year: 2003
keywords: [oil prices, GDP, nonlinearity, energy expenditure share, de-dollarization, staples]
pdf_url: "https://econweb.ucsd.edu/~jhamilton/oilshock.pdf"
---
## Abstract

Hamilton's canonical treatment of the oil–macroeconomy relationship. Argues that oil price increases matter for real GDP only insofar as they exceed the maximum price seen in the previous several years (the "net oil price increase" measure), and that oil price *declines* do not matter — the relationship is fundamentally nonlinear. The paper also frames the *economic significance* of oil in terms of **oil expenditure as a share of GDP**: energy's real cost burden is measured not by the nominal $/bbl quote but by what share of national output it consumes. That is the intellectual basis of the "de-dollarized staple cost" framing: a $100 barrel means something entirely different at $20T GDP than at $2T GDP.

## Why it's in the StratMac library

- Provides the theoretical justification for the `*_GDP` staple cards: price ÷ GDP strips money illusion and measures the real output share a staple claims.
- The "expenditure share" variant (price × quantity ÷ GDP) is the middle-choice rescale for bringing de-dollarized ratios back to recognizable numbers (~2–3% of GDP for oil today vs ~8% in 1980).
- The net-oil-price-increase nonlinearity is a candidate refinement for the GAS/WTI/BRENT sigma signals.