Alternative inflation estimates attract attention because household experience does not always resemble a headline index. The central question is not whether the consumer price index has limitations—it does—but whether a proposed alternative can survive transparent, independent checks.
What the alternative critique gets right
The official consumer price index is not a literal price tag for one permanently fixed basket. Statistical agencies account for changes in what households buy, the arrival and disappearance of products, and measurable changes in quality. That makes the index more useful for some economic questions, but it can also move it away from the intuitive idea of pricing exactly the same shopping basket year after year.
Public perceptions can differ from published inflation for understandable reasons. Households buy different things, face different housing and medical expenses, and often notice large or frequent price rises more readily than stable prices. A national average therefore cannot describe every household's experience.
Two different questions
A fixed-basket index asks how much the same items cost now. A cost-of-living index asks how much expenditure is needed to maintain a comparable level of satisfaction when prices and purchasing patterns change. Neither question is meaningless, but they are not interchangeable.
Substitution
Suppose the price of one food rises sharply while a close alternative remains stable. A fixed-basket measure prices the original quantities. A cost-of-living approach recognises that many consumers will change the mix. Critics see this as understating the loss involved in giving up a preferred item; economists respond that ignoring the change can overstate the money required to reach a comparable level of welfare.
Quality adjustment
Products also change. A tyre that lasts twice as long, or a computer with substantially greater capability, is not identical to its predecessor. Failing to recognise useful improvements can overstate inflation. The difficult part is determining how much value consumers place on each improvement, especially when new capabilities matter greatly to some buyers and little to others.
Housing and medical care introduce further challenges because the amount paid directly by a household does not always match the economic value statistical agencies are trying to measure. These are valid reasons to read an index with care rather than treat it as a perfect account of every budget.
Where the large alternative estimate runs into trouble
The disputed series argued that methodological changes had reduced reported annual inflation by a very large and growing margin. If compounded over decades, that assumption implies a dramatically greater increase in the price level than the official index. The original analysis tested that implication against information not dependent on the inflation index itself.
Grocery-price crosscheck
Historical newspaper advertisements provide an imperfect but intuitive benchmark. The 2015 exercise compared selected grocery prices from a 1982 advertisement with then-current shelf prices, adjusting for package size. The official index was not exact, but its implied prices were substantially closer to the observed examples than those generated by the larger alternative adjustment.
This was not a definitive inflation index: sale prices, local markets and a small sample limit the exercise. Its value was diagnostic. A series claiming a very large cumulative rise should produce plausible estimates for ordinary goods. In this comparison, it did not.
Catalogue-price crosscheck
Old retail catalogues allow a second test. Televisions, appliances and other goods often became less expensive relative to their capabilities. Even without fully valuing quality improvements, the large alternative inflation adjustment implied modern prices far above prices actually observed in the 2015 market. Lower-technology products produced the same broad concern.
Physical-output crosscheck
Applying a much higher inflation rate to nominal output also creates a much weaker estimate of real economic growth. The archive analysis compared that implication with physical indicators such as vehicle sales, freight movement and electricity generation. Those indicators had limitations of their own, but their broad growth was difficult to reconcile with an economy supposedly contracting in most years.
The conventional real-output series offered a more coherent explanation: services could grow faster than goods, freight could lag total output, and efficiency gains could allow electricity use to rise more slowly than the economy.
Interest-rate crosscheck
Nominal interest rates usually include compensation for expected inflation. Under the Fisher-effect framework, subtracting an inflation estimate gives an approximate real rate. Using the much higher alternative inflation series implied persistently negative real mortgage rates over a long period, meaning lenders would repeatedly accept substantial losses in purchasing-power terms.
Real rates can be negative, expectations can be wrong and markets are not perfectly stable. The concern was persistence and scale: the alternative interpretation required an implausible pattern rather than an occasional departure.
The methodological issue
The main criticism centred on a comparison between the published CPI-U and a research series designed to estimate how earlier data might look if later methods had already been in place. Comparing the two can help estimate the effects of methodology changes. The dispute concerns how those annual differences should be combined.
The alternative calculation treated successive annual differentials as quantities to be added into a running total. The archive analysis argued that this double-counted later methodological changes because an earlier differential already reflected changes that had not yet entered the contemporaneous published series.
A simplified illustration
If one methodological change reduces measured inflation by 0.8 percentage points and a later change reduces it by another 0.5 points, their permanent combined effect is 1.3 points. Adding the full gap again in each intervening year would repeatedly count the same changes.
On this interpretation, the better diagnostic is the gap between the research series and published series before the relevant changes took effect, allowing for year-to-year noise. The original article estimated that the portion attributed to those documented changes was much smaller than the running-total approach suggested.
Bottom line
- The official CPI is a constructed national measure, not a perfect description of every household's expenses.
- Substitution, quality, housing and healthcare deserve careful explanation and continued scrutiny.
- An alternative index must produce plausible results when checked against independent prices, output and financial-market behaviour.
- The running-total method examined here appears to count some methodological effects more than once.
- Disagreement about statistical design is not, by itself, evidence that source data were falsified.
The search for useful supplementary inflation indicators remains worthwhile. Household-specific measures, distributional analysis and clearly defined fixed-basket indexes can all add context. Their assumptions should be published, their calculations reproducible and their results tested against observable evidence.
Continue with part two
The follow-up article applies the same evidence-first approach to an alternative unemployment measure and asks what different labour-market indicators actually capture.