The latest inflation report from the U.S. Department of Labor showed an acceleration in price increases, with the consumer-price index (CPI) rising 0.6% from December 2021 to January 2022. The annual increase, measured without seasonal adjustments, was 7.5% — a four-decade high. As troubling as these numbers are, the truly scary figure is hiding behind them.

A recent change in the computation of the CPI by the Bureau of Labor Statistics (BLS) is downplaying inflation. In reality, the general level of prices rose 0.7% from December to January, not 0.6%. That may not sound like much, but it is the difference between an annual rate of 7.4% and 8.7% — a 1.3 percentage point deviation.

The divergence between these two numbers is due to a process called reweighting. The CPI functions like a weighted average of all the goods and services which typical urban consumers purchase, but it needs to be periodically updated to reflect consumers’ changing preferences. That is done every two years through changing the weights assigned to the components of goods and services within the index and is part of the normal procedure at BLS.

The December 2021 inflation report was the last month to use data from 2017 to 2018 to weigh the components of the CPI. Beginning with this latest inflation report, the weights are now derived using data from 2019 to 2020. Therein lies the problem.

In 2020, consumers’ behavior changed dramatically, and they had an abnormal basket of goods and services, largely due to government-imposed lockdowns and fears of the COVID-19 pandemic. People bought less gasoline and fewer vehicles than normal, and that reduced the new weights assigned to those items in the CPI. Coincidentally, these are also categories with prices which have been rising faster than the overall price level.

Conversely, people spent more of their budgets than normal on medical care commodities, which increased the new weight assigned to that category. But the price of this component of the CPI has been growing slower than the overall index.

In short, more weight has been placed on portions with slower-growing prices in the CPI and less weight has been placed on categories with faster-growing prices. The result is an underestimation of inflation. If the previous weights are applied to the new price increases, I estimate the monthly rise in the CPI is 0.7%, not 0.6%, for an annualized rate of 8.7%, not 7.4%. Additionally, real wages were flat instead of increasing 0.1%.

There likely was some concern at the BLS that this problem would arise; its website noted that the agency considered altering its normal procedure but decided against it.

In December, I inquired at BLS as to who made the decision to follow the normal procedure despite the anomalous data in 2020. Its response did not name a particular individual or even a committee, but said “We ultimately decided the costs outweighed the benefits.”

The issue is not that the weights were adjusted — that is a sensible, normal process and has been the practice for years. The problem this time is that the BLS is using anomalous data that is likely a less-than-ideal representation of the typical urban consumer’s basket of goods and services today.

Eerily, it seems a sign of the times. This underestimation of inflation fits in well with an age of Orwellian newspeak, wherein the White House first denied there was inflation, then said inflation was transitory, then said inflation was good. One consequence of a lower headline number will be to provide some modicum of political cover to an administration which has been dogged by falling poll numbers in its handling of the economy.

But there is no numbers game that can hide the reality facing American consumers. The veil of an artificially low headline CPI number does not diminish the pain of filling a gas tank or buying groceries. It is only by virtue of a mathematical fluke that inflation appears lower than it really is. But if one peers at the formula behind the curtain, the illusion fades away and the disappointing reality is apparent.