Inflation Rate from a Price Index

An inflation rate is a percentage change in a price index, and everything difficult about it is in the index rather than in the arithmetic. The division itself has exactly one trap worth naming: divide by the EARLIER index, always. Going from 137.5 to 141.9, the rate is 4.4/137.5 = 3.20%. Dividing by 141.9 instead gives 3.10%, and the gap widens as the rate rises. The rate is measured against where prices started, because that is what a percentage change means.

What a price index is, is a weighted average of the prices of a basket of goods, rebased so some year reads 100. Everything contentious lives in the word "basket". The index measures the average price of a specific bundle bought by a specific notional household, and no actual household buys that bundle. A young renter in a city and a retired homeowner in a small town face different inflation rates in the same year, sometimes by several percentage points, because rent and health care and fuel carry different weights in their lives than in the published basket. Both figures are correct; they are answers to different questions.

Three technical problems make the number genuinely hard, and statistical agencies publish long papers about all three. Substitution: when beef gets dear people buy chicken, and a fixed basket keeps pricing the beef, overstating the cost of living. Chained indices attempt to correct this and are one reason two published series diverge. Quality change: a car today is not the car of 1980 at a higher price, and separating the price rise from the quality rise is a judgement call made by statisticians using methods reasonable people disagree about. New goods: something that did not exist enters the basket only after it becomes common, so the enormous price falls of its early years are never recorded at all.

Two practical notes on using the rate you compute. First, an annual rate computed from monthly index values needs the twelve-month comparison, not the one-month change multiplied by twelve, because month-to-month movements carry seasonal noise that swamps the trend. Second, inflation rates do not add across periods — 3% then 3% is 6.09%, not 6% — so a multi-year rate is a compounding calculation and treating it as a sum drifts noticeably past about five years.

Finally, the distinction that causes the most argument: inflation is a change in the general price level, not in any one price. Rent rising while electronics fall is a change in relative prices, and it can happen at any inflation rate including zero. When published inflation feels wrong against your own experience, the usual explanation is that your basket is not the average basket — and that is a fact about weights, not evidence that the index is fabricated.