What Really Goes Into the Fed's Favorite Measure of Inflation?

Episode Summary

Executive Summary: The episode explains why inflation is measured differently by CPI and PCE, why the Fed prefers PCE, and how the underlying data are actually assembled. Guests Omer Sharif and Skanda Amarnath show that methodology, weighting, and imputation can produce big real-time divergences—especially in services, insurance, airfares, and financial services—making inflation figures more subjective and more consequential than they first appear.

Main Topics: CPI vs. PCE: different scopes and purposes (Priority: 5/5): The guests explain that CPI is more focused on consumers’ out-of-pocket spending, while PCE is broader and includes third-party or government-paid items such as Medicaid. This scope difference is the main reason the Fed prefers PCE. Why the Fed shifted toward PCE (Priority: 5/5): Skanda provides historical context: the Fed moved from mainly referencing CPI to PCE around 2000, driven by concerns about substitution bias, quality change, and a desire for a more dynamic, representative index. Current wedge between core CPI and core PCE (Priority: 5/5): The discussion highlights that the spread between core CPI and core PCE has widened sharply versus historical norms, with core PCE slowing faster than core CPI and the divergence now around a full percentage point. How inflation data are actually collected (Priority: 4/5): The conversation walks through how CPI uses consumer expenditure surveys, price collection across metro areas, and detailed basket construction, underscoring the large role of survey design and response quality. Imputed and hard-to-measure categories (Priority: 5/5): Guests discuss items like owner’s equivalent rent, financial services without explicit payment, and portfolio management fees, showing how PCE and CPI rely on proxies and imputations rather than direct prices. Goods vs. services is often a false divide (Priority: 4/5): Skanda argues that many 'services' categories are heavily influenced by goods prices and supply shocks, using airfares and auto insurance as examples where commodity and parts costs matter as much as wages. Timing, seasonality, and market relevance (Priority: 4/5): The episode covers January residual seasonality, why CPI often moves first, and how inflation numbers affect markets, Social Security COLAs, leases, and Fed expectations before the PCE release arrives.

Key Arguments: CPI and PCE are designed to measure different scopes of inflation, not merely the same thing in two formats; PCE captures broader economic activity and third-party payments. The Fed prefers PCE because it is more representative of the economy and adjusts more dynamically to changing consumer spending patterns. Core CPI has historically run above core PCE by about 30 to 50 basis points, but the spread has recently widened to around 100 basis points because core PCE has been slowing faster. Many inflation categories depend on imputations and proxies, so the numbers are not purely mechanical reflections of observed sticker prices. Owner’s equivalent rent is a key conceptual difference, but it is not the only important one; health care, insurance, airfares, and financial services also diverge materially across indexes. Survey nonresponse is a growing issue for inflation measurement, but the BLS and BEA still have robust methodologies and are responsive to technical questions. Services inflation is often linked to goods markets and supply conditions, not just wages, especially in categories like auto insurance and airfares. Measured financial services in PCE can move with equity-market performance because portfolio management fees are estimated from returns reported by institutions. January price increases often reflect residual seasonality, but this year the strength was unusually concentrated in services rather than core goods.

Data Points: Typical core CPI vs. core PCE spread: 30 to 50 basis points - Historical year-over-year spread between core CPI and core PCE before the pandemic, with CPI higher Current core CPI (year-over-year): 3.9% - Referenced as the latest core CPI reading during the discussion Current core PCE (year-over-year): About 2.8% to 2.9% - Guests estimated core PCE well below core CPI, implying a roughly 1 percentage point gap Shelter weight in CPI: About 43% - OER/shelter is much heavier in CPI than in PCE Shelter weight in PCE: About 15% to 16% - Smaller shelter influence in PCE than in CPI Core PCE share from imputed prices: About 13% - Omer noted a sizable share of core PCE consists of imputed prices not directly observed Auto insurance inflation in CPI: About 21% y/y - Example of a category running much hotter in CPI than in PCE Auto insurance inflation in PCE: About 8.7% y/y - Same category measured very differently in PCE Portfolio management services contribution: About 8 basis points - Expected contribution to January core PCE from equity-market gains flowing through measured financial services S&P 500 Q4 performance: About 12% to 13% higher - Referenced as the driver behind stronger portfolio management fee estimates Market-based core PCE weight of portfolio management services: About 1.5% - Used to explain why strong equity returns can still materially affect PCE January effect / residual seasonality: First quarter usually stronger; two-thirds of the strength is in core goods - Pattern in CPI seasonality, though this year strength was unusually concentrated in services Auto insurance annual print example in PCE vs CPI: CPI 21% vs. PCE 8.7% - Presented as an illustration of how methodology changes growth rates drastically Core PCE forecast: 0.36% month over month - Both guests gave this estimate for the upcoming release Super core PCE forecast: 0.50% month over month - Both guests expected a stronger read in super core PCE

Pivotal Quotes: "If you exclude everything that you need to live, then inflation is coming down." — Joe Weisenthal: A humorous critique of cherry-picking inflation measures that confirm one’s prior beliefs "The PCE is a much broader index. It captures essentially more of the economy... than what the CPI does." — Omer Sharif: Core explanation of why PCE and CPI can diverge materially "Roughly 13% of the entire core PCE is just these imputed prices that no one sees." — Omer Sharif: Used to emphasize how much of PCE depends on indirect measurement and proxies

Implications: Listeners should treat inflation readings as constructed estimates, not pure facts. Methodology differences can materially change Fed policy expectations, market pricing, COLAs, and inflation-linked investments.