This Week In The Economy: Structural Problems In The Consumer Price Index

Welcome to a regular snapshot-review of U.S. and international economic news that aims to 1) provide a window into the challenges and decisions facing businesses today, 2) determine the direction of economic policy — such as the speed at which central banks decide to raise interest rates, and 3) assess what the impact will be for consumers.

Since the COVID-19 pandemic, inflation has dominated economic conversations — with affordability and rising cost of living issues plaguing the Biden and Trump Administrations, casting a cloud over the legacy of former Federal Reserve Chair Jerome Powell, and complicating his successor’s planned agenda.

In this edition of TWE, let’s examine one of the key indicators used to measure the inflation, the Bureau of Labor Statistics’ Consumer Price Index — of vital importance because besides being a measure of the economy’s health, the CPI is used to adjust income eligibility levels for government assistance, Social Security checks, federal tax brackets, and wages tied to collective bargaining agreements. It also influences the Fed’s interest rate decisions (even though it is not the central bank’s preferred inflation gauge).

In particular, we are going to highlight the structural problems with CPI — issues that, despite attempted fixes by the BLS in recent years, remain. This results in an indicator that sometimes tells a completely different story on inflation from what most consumers are actually experiencing.

The Headline CPI Number Is Dominated By Three Imputed Series

Shelter inflation lags new-tenant rent growth by three to four quarters due to tenant composition dynamics — specifically because the CPI captures rents paid by all tenants (new and existing leases), not just current market transactions.

It takes nearly two years for contract rents to catch up with market rents, reflecting the typical term of a lease and the fact that landlords do not necessarily raise rent immediately to full market rate upon renewal.

The consequence: excluding shelter, core CPI (meaning excluding food and energy prices) has been growing at about 2% on a year-over-year basis since mid-2023 — a number that would have substantially altered the monetary policy conversation.

Owners’ Equivalent Rent is a Survey-based Guess

Within the shelter category, OER is the single largest line item. OER is based on owner-occupied housing and collected by asking homeowners: “If someone were to rent your home today, how much do you think it would rent for monthly, unfurnished and without utilities?”

The CPI program collects rent data from each sampled unit every 6 months, and many rents change infrequently, being locked in place for a given lease term. The published monthly figure is an average of the last six months of these surveys — meaning any given month’s survey counts for just one-sixth of the published estimate. The execution introduces substantial noise and lag.

Medical Costs Are Calculated Using An Indirect, Counterintuitive Method

The CPI for health insurance measures the price of health insurance services directly provided by insurance companies, but it excludes the price of indirectly purchased medical goods and services, which make up the majority of total premiums.

Instead of tracking what people actually pay in premiums, BLS uses an “indirect method” based on insurer retained earnings. Actual premium and claims costs are significantly higher than the Medical Care CPI has recorded. The CPI does not capture rising administrative costs, from billing complexity to regulatory compliance, which can inflate premiums even if actual service costs remain stable.

NOTE: BLS updated this method in October 2023 to use a rolling two-year average, which reduced volatility but left the conceptual problem intact.

The Digital economy is Chronically Underweighted

Official CPI inflation numbers tend to overstate the true rate of inflation due to various problems such as substitution bias in the fixed-weight index and failure to account adequately for quality change — with the digital economy a major source of that gap.

The shift from cable TV to streaming, landlines to mobile, and physical media to cloud access represent enormous price-quality changes that hedonic models only partially capture. Free services (Google, Facebook, navigation apps) are excluded entirely by design, since the CPI only tracks direct consumer expenditure.

Bottom Line

The CPI is methodologically sound in its basic design, and BLS has made meaningful updates in recent years, for example used car mileage adjustments in 2024, physician services claims data in November 2024, and health insurance smoothing in 2023).

But the index has three deep structural problems that are hard to fix without redesigning it:

• The shelter/OER lag amplifies housing cycles into the headline number with a 1–2 year delay.
• Health insurance cost measurement is conceptually backward;
• The digital economy’s growing share of consumer welfare is either underweighted or entirely outside scope.

The result is an index that policymakers and markets treat as a precise read on consumer prices, but which can diverge substantially from what consumers are actually experiencing in real time.