Inflation Base Effects: Why Year-Over-Year Rates Can Fall While Prices Rise

Year-over-year inflation compares a price index with its value in the same month a year earlier. The comparison is relative to the older index level, so the size of the price change recorded in that comparison month affects today’s annual rate. If a large increase from a year ago moves out of the 12-month window, the annual rate may ease, all else equal. If a small increase or decline moves out, the annual rate may rise, all else equal.

The phrase “drops out” is shorthand. The old observation is not deleted from the historical index. The comparison month rolls forward: the current index is compared with a newer value from 12 months earlier. The Bureau of Labor Statistics (BLS) uses this kind of example to explain why a year-over-year CPI rate can fall while the latest monthly index still increases. Read its explanation of the math behind recent inflation trends.

This mechanism can affect CPI, PCE, or another price index calculated over a rolling year. The details depend on the index, its weights, and its monthly observations. A base effect does not mean statisticians changed the basket or adjusted the index incorrectly; it describes how the comparison period influences a rate of change.

Let P(t) be the price index for the current month. Its year-over-year inflation rate is:

(P(t) ÷ P(t−12) − 1) × 100

The next month’s year-over-year rate compares P(t+1) with P(t−11). Both the current index and the comparison value have moved forward by one month. The earlier comparison-month movement therefore matters along with the newest monthly change.

For one consistent index series, the relationship between adjacent annual rates can be written using monthly growth rates:

1 + annual rate now = (1 + annual rate last month) × (1 + current monthly change) ÷ (1 + monthly change one year earlier)

Rates in this relationship are expressed as decimals. This identity explains why the annual rate can change even when the latest month has a positive change: it depends on whether the current monthly increase was larger or smaller than the increase in the month that is leaving the comparison window. The BLS explains that 12-month CPI changes compare the same month across years and should not be confused with the sum of the intervening monthly changes ({source:blsCpiPercentChanges}).

A 12-month rate also summarizes all the monthly changes inside its window; it does not isolate the most recent month. For a quick reading, look at both the year-over-year figure and a suitable recent monthly measure. They answer different questions.

Suppose a fictional price index showed a 6.0% year-over-year increase last month. This month, the index rises another 0.3%. But the monthly increase from the same month a year ago was 1.2%. The new annual comparison is approximately:

1.06 × 1.003 ÷ 1.012 − 1 = 0.0506, or about 5.06%

The year-over-year rate falls from 6.0% to about 5.06%, a decline of roughly 0.94 percentage points. Yet the current index rose 0.3% this month. The annual rate eased because the smaller current monthly increase replaced a larger increase from a year earlier in the rolling comparison.

All inputs are hypothetical and are included only to show the arithmetic. They are not current CPI or PCE readings, a forecast, or evidence that any particular economy’s inflation is easing. The BLS has illustrated the same kind of arithmetic with historical CPI data: a positive latest monthly change can coincide with a lower annual rate when a larger year-earlier monthly change leaves the window. Its [guide to calculating CPI percent changes]({source:blsCpiPercentChanges}) explains how to calculate one-month and 12-month changes.

<!-- learn:illustration -->

The effect depends on the old monthly move that leaves the window relative to the new monthly move that enters it. If the current month’s increase is smaller, the 12-month rate tends to fall, all else equal. If the current increase is larger, the annual rate tends to rise, all else equal. A past price decline leaving the window can also lift the annual rate even if the latest increase is modest.

This is why an annual inflation reading can turn upward after a temporary dip without a sudden acceleration in the latest month. The opposite can happen too: a prior-year spike can leave the window and pull the annual rate down even though current monthly increases remain positive. The St. Louis Fed’s explanation of [falling inflation rates and rising prices]({source:stlouisPriceLevelInflationRate}) separates the direction of the price level from the speed at which it changes.

The arithmetic can indicate how much the comparison base is influencing the change in the annual rate, but it cannot show that current price pressure has disappeared. For that, readers need to examine current monthly changes, a few months of trend, index components and their weights, and the specific measure being used. The rolling comparison is one part of the evidence, not a verdict on underlying inflation.

“Base effect” and “base period” sound similar but refer to different things. Many CPI index series are normalized to equal 100 in a reference period. That reference helps express the index level; it is not the old comparison month that rolls out of a year-over-year rate. BLS documents the CPI’s index values and reference periods in its [CPI concepts handbook]({source:blsCpiConcepts}).

Seasonal adjustment is different again. It estimates recurring patterns within the year, such as usual holiday or weather effects, to make nearby periods easier to compare. A year-over-year rate compares a month with the same month a year earlier; this often reduces recurring seasonal patterns, but does not necessarily remove every calendar, unusual-event, or measurement effect. Check whether the published monthly value is seasonally adjusted or not before comparing it with another series. See the guide to seasonally adjusted and unadjusted data.

A base effect is also separate from inflation adjustment. It does not convert nominal dollars into constant purchasing-power terms. It is about how the starting value in a percentage comparison affects the reported rate.

A broad inflation index combines many prices, and categories do not contribute equally. A sharp year-earlier move in a heavily weighted category can affect the aggregate more than a similar percentage move in a small category. The BLS CPI and the Bureau of Economic Analysis (BEA) PCE price index have different coverage, weights, and calculation details, so their measured base effects need not match ({source:beaCpiPceComparison}). The BEA describes what the [PCE price index measures]({source:beaPcePriceIndex}).

Volatile categories such as energy can produce noticeable changes in annual comparisons, but a base effect is not limited to energy. It can arise in any index where a large or unusual monthly price change becomes part of the year-earlier comparison. Tax changes, temporary discounts, rebates, supply disruptions, or unusually large price moves may affect a category’s path; their effects on the total index depend on the measure and weights.

Do not compare the raw index levels of CPI and PCE as if one index were a price tag for the same basket. Instead, name the series and compare percentage changes over matching periods. For the broader scope differences among CPI, PCE, and the GDP deflator, see CPI vs. PCE vs. the GDP deflator.

A base effect is useful for understanding why an annual rate moved from one month to the next. It helps explain the arithmetic behind statements such as “inflation fell even though prices rose.” It does not mean the index’s historical values changed, and it does not establish that the current monthly pace is low.

A base effect is not itself a forecast. The next year-over-year reading will depend on future monthly price changes and on the comparison month that will leave the window then. A calendar of known comparison-month changes can help explain why the arithmetic may push a rate up or down, but it cannot guarantee the result because the new data are not yet known. Keep projected base effects separate from projections of future prices.

Nor does a lower annual rate automatically mean that households are paying less than a year ago. If the price index remains above its year-earlier level, prices in that measured basket have risen over the 12-month period, even if the rate of increase slowed. For the broader distinction between a slower positive rate and falling prices, see disinflation vs. deflation.

Start with the name of the index and the period being compared. Is the release reporting CPI or PCE? Is the headline month-over-month, year-over-year, or an annual average? The BLS notes that a December-to-December change is not the same as the change in the annual average for those two years ({source:blsCpiPercentChanges}).

Then inspect both the index level and the rate when possible. A positive monthly change with a lower annual rate is not contradictory: the latest index can rise while the comparison month from a year earlier had a larger increase. If the rate is seasonally adjusted, check its label and compare it with a series calculated on the same basis. Do not treat an annual rate as if it were a direct measure of only the latest month.

Finally, distinguish the numerical explanation from an economic conclusion. A base effect can explain why a 12-month rate shifted; it cannot by itself establish what caused the latest price changes or whether inflation has returned to a particular goal. State the measure, period, monthly direction, annual-rate direction, and relevant year-earlier comparison. That makes the headline easier to verify and avoids saying that prices fell when only the rate declined.