Omair Sharif Explains How Inflation Measures Really Work

Episode Summary

Executive Summary: The episode argues that inflation is best understood through granular, bottom-up analysis rather than broad macro theories. Guest Omer Sharif says economists lack a cogent, durable theory of inflation, so forecasting depends on component-level details, index construction, seasonality, and timing. He explains why CPI and PCE differ, why shelter lags market rents, why some categories are still puzzling, and why current inflation may cool as transitory goods shocks fade, even if some wage and shelter pressures persist.

Main Topics: Inflation theory is weak and regime-dependent (Priority: 5/5): The hosts and Omer Sharif agree that economists do not have a universally reliable theory of inflation. Relationships like the Phillips curve may work in some periods but fail in others, so forecasting depends heavily on the current economic regime. Bottom-up inflation forecasting (Priority: 5/5): Sharif argues that detailed, component-by-component analysis of CPI/PCE often beats top-down models. He emphasizes persistence, seasonality, and index mechanics over abstract macro narratives. Which inflation measures matter (Priority: 4/5): The discussion contrasts CPI and core PCE, explaining that the Fed focuses on core PCE while markets often care more about CPI because of TIPS pricing and trading relevance. Shelter, OER, and lagged housing inflation (Priority: 5/5): The episode explains why owners’ equivalent rent and shelter measures lag private rent data and home prices, due to both conceptual choices and slow sample turnover in CPI. What is transitory inflation? (Priority: 5/5): Sharif defines transitory inflation as a temporary acceleration in the rate of change, not merely a high level. He expects goods-driven spikes like used cars to fade, though he watches for wage pass-through. Methodology, seasonality, and hard-to-forecast categories (Priority: 4/5): The conversation highlights that some inflation movements are driven by statistical construction and seasonal factors, with apparel and airfares cited as especially difficult categories to forecast. Investors’ use of inflation research (Priority: 3/5): Sharif describes serving institutional clients with timely, actionable forecasts and detailed component analysis that can help fixed-income and TIPS traders anticipate prints before official releases.

Key Arguments: Economists do not have a cogent, stable theory of inflation; what works changes by cycle and decade. Simple naive models, such as averaging the last four quarters, can outperform more sophisticated Phillips curve models in many periods. Inflation persistence is crucial, but it varies over time, which is why some models succeed in one decade and fail in the next. The Fed is less dependent on the Phillips curve than it used to be and instead uses a suite of measures including core inflation, trim means, and medians. CPI and core PCE are the two main U.S. inflation measures; the Fed uses core PCE, while markets care more about CPI because it drives TIPS pricing. Headline inflation numbers are useful constructs, but some components—like energy and cars—better reflect reality than debated items like OER. Owners’ equivalent rent was adopted because the old asset-based housing treatment created too much volatility and mixed investment with consumption. Private rent and home-price indices move faster than CPI shelter because CPI has limited turnover and the shelter component lags 12 to 18 months. Current shelter inflation may rise, but expected increases are likely to be offset by declines in used cars and other transitory components. Breadth of inflation has not dramatically worsened; many components are rising, but the share is only modestly above pre-pandemic norms. Broad macro stories like supply chains matter, but index mechanics, seasonality, and measurement details often explain more of the monthly move than narratives do. Wage pass-through into services, especially restaurants and hotels, is one of the key risks that could make inflation more persistent. Some categories, especially apparel and airfares, remain extremely difficult to forecast even with alternative data sources. The bottom-up approach is most valuable for institutional investors who need to understand what is driving the next print and whether the move will persist.

Data Points: CPI inflation: about 5.4% - Joe notes the headline CPI is a little over 5%, down from a recent high, during the discussion of current inflation levels. Fed forecast horizon: about 3 years - Sharif says the Fed’s forecast horizon is essentially three years, making decade-scale inflation regime changes less relevant to policy models. Core CPI historical range: 1.5% to 2.5% - Sharif uses this range as an example of why naive models can work well when inflation has been stable for years. Rent and OER combined: about 2.5% - Sharif says rent plus owners’ equivalent rent is around 2.5% now, versus around 3.5% pre-pandemic. Potential shelter contribution to core CPI: about 80 basis points - He estimates that if rent and OER rose 2 percentage points, it would add roughly 0.8 percentage points to core CPI. Used cars contribution to core CPI: over 130 basis points - He says used cars are currently adding more than 130 bps to core CPI year over year. Share of PCE components rising: about 84% to 85% - Sharif cites San Francisco Fed tracking showing most PCE components currently have price gains, only modestly above pre-pandemic levels. Core PCE forecast: around 2% by mid-2022 - Sharif expects core PCE to slow toward 2% by the middle of 2022. Core CPI forecast: around 2.5% by mid-2022 - He expects core CPI to remain somewhat higher than core PCE. Share of CPI with quality adjustments: about 4% to 5% - He notes hedonic adjustments apply only to a small part of CPI, mostly goods. Wireless index move: -7% in one month - He cites the March 2017 unlimited wireless plan change as a major hedonic-adjustment-driven decline in the wireless index. Impact on core CPI from wireless: just over 0.1 percentage point - The wireless adjustment subtracted a bit over a tenth from monthly core CPI. BLS CPI inputs: 211 indicators and over 7,000 basic item-area indexes - Sharif describes the granularity of the CPI construction process. Monthly CPI sampling turnover: 10% to 15% - He says only a small portion of the rent sample represents new renters each month, explaining shelter lag. Class A city rent weight: 20% from New York, Los Angeles, and Chicago combined - He notes these three large metros carry a disproportionate share of the rent index. Core CPI monthly change: around 0.3% - Used to illustrate that a 2.5% monthly drop in motor vehicle insurance is large relative to the core monthly print. Motor vehicle insurance: down about 2.5% last month - Sharif attributes this to seasonality rather than a macro story. Airfares vs trend: 13% below expected trend - He says airfares are still below their pre-pandemic trend, suggesting room for some normalization upward. Hotel rates vs trend: 8% above expected trend - He cites hotel rates as an area where prices are above trend and may give back. Rent weight in core CPI: about 40% - Used in his estimate of how rent increases translate into overall core inflation. Wholesale used vehicle prices: down 3% to 4% over recent months - He cites recent declines as inventories normalize. Pre-pandemic rent pace: about 3.5% - Used as a comparison point for current rent/OER growth.

Pivotal Quotes: "I think the short answer to that is no." — Omer Sharif: Answering whether economists have a cogent theory of inflation. "All of this stuff is a construct, right?" — Omer Sharif: Explaining that inflation indices are useful but methodologically constructed measures, not pure reality. "inflation is always and everywhere a monetary phenomenon" — Joe Weisenthal: Invoking Milton Friedman while asking how monetary policy fits into Sharif’s bottom-up framework.

Implications: Listeners should expect inflation forecasts to hinge on component detail, not slogans. For markets and the Fed, shelter lags, wage pass-through, and category-specific quirks matter more than broad narratives, suggesting inflation may cool unevenly rather than in a straight line.