CPI, Bins, Policy, Inflation
Every month, the Bureau of Labor Statistics (BLS) releases the Consumer Price Index (CPI), and markets hold their breath. The number that emerges, a single, authoritative figure like 3.5%, is treated as an objective measure of reality. Central bankers at the Federal Reserve and the Bank of England adjust interest rates, governments calculate cost-of-living adjustments, and fortunes are won and lost based on its subtle movements.
But what if this number is less a reflection of reality and more of a spreadsheet artifact? What if it's a carefully constructed illusion, a histogram whose final shape is determined by choices that are as much political as they are statistical?
The dirty secret of the CPI is that its ultimate value can be swung by half a percentage point or more simply by changing the weights of its constituent categories. This isn't a conspiracy theory; it’s a methodological reality. And it raises a disquieting question: are central banks targeting inflation, or are they targeting a statistical construct of their own choosing?
The Inflation You Feel vs. The Inflation They Report
The CPI is not a simple average. It's a weighted average of price changes across a basket of goods and services, from gasoline and groceries to rent and used cars. The core of the issue lies in how these weights are determined. The BLS uses consumer expenditure surveys to figure out what the "average" household buys. As of recent data, "Shelter" is the titan, accounting for over a third of the entire index. "Food" is around 13%, and "Apparel" is a mere 2.5%.
Here’s where it gets interesting. These weights change over time, and the methodology for measuring the prices within them is constantly being refined. Consider the three titans of modern consumer spending: Shelter, Food, and Technology.
Shelter: The cost of shelter is notoriously difficult to measure. The BLS primarily uses a concept called "Owners' Equivalent Rent" (OER), which asks homeowners what they think they could rent their house for. This is a survey-based, lagging, and somewhat imaginary number. When house prices soar, OER moves with a significant delay. This choice alone smooths out and delays the biggest driver of inflation for most families.
Food: Food prices are volatile. A drought, a war, or a supply chain snarl can cause prices to spike dramatically. These are real, painful costs that consumers feel immediately.
Technology & Quality Adjustments: The price of a new iPhone might be the same as last year's model, but the BLS will record a price decrease because the new phone is "better" (faster processor, better camera). These "hedonic quality adjustments" systematically introduce a deflationary force into the index. The Bank of England has published blog posts wrestling with this very measurement bias, acknowledging that it’s hard to disentangle true price changes from technological progress.
The 0.5% Illusion
Now, let's run a thought experiment. Imagine a year where shelter costs (as measured by the slow-moving OER) rise by a modest 4%. Meanwhile, food prices, battered by supply shocks, jump 10%. And the "quality-adjusted" price of electronics and software falls by 15%.
Under the current weighting, the heavy-handed influence of shelter keeps the headline CPI number tame. But what if we used a different, equally plausible weighting system? What if we weighted categories based on their volatility or their importance to lower-income households, who spend a much larger percentage of their income on food and gas?
By slightly increasing the weight of food to reflect its outsized impact on household budgets and slightly decreasing the weight of tech (arguing that a new phone is a discretionary purchase, not a core living cost), you can easily swing the final headline CPI number by more than 0.5 percentage points.
In Scenario A (Official Weighting), the low-drama shelter number anchors the index, and the headline inflation reads 3.2%. The Fed breathes a sigh of relief.
In Scenario B (Alternative Weighting), the painful spike in food prices is given more prominence, and the headline inflation reads 3.8%. The Fed is now behind the curve and pressured to hike rates.
Which number is "real"? Both are. They are simply different histograms drawn from the same raw data. The choice of bins and weights determines the story.
Targeting a Spreadsheet
This is not just an academic exercise. When a central bank has a mandate to target 2% inflation, it is targeting 2% as defined by the CPI's specific methodology. They are not targeting the pain at the pump or the shock at the grocery store. They are targeting a weighted average that is heavily influenced by a lagging, theoretical rent figure and systematically lowered by questionable technological adjustments.
The process creates a strange feedback loop. The institution responsible for fighting inflation is also influenced by the methodology of the very index it is supposed to control. This is not to say the statisticians at the BLS are acting in bad faith; they are grappling with an impossible task. But the result is a single number that launders a messy, complex reality into a clean, actionable, and potentially misleading data point.
The next time you see the monthly CPI release, don't ask, "What is inflation?" Ask, "How did they choose to calculate it this month?" The answer reveals more about the map than it does about the territory.
