Consumer Price Index Calculator

What the index number measures

The consumer price index is not a price. It is a ratio, built from the cost of a fixed basket of goods and services, and it is published against a base period that is set to 100. For the CPI-U the base is the average of 1982, 1983 and 1984. An index reading of 321.9 in 2025 means the same basket that cost $100 in the base period now costs $321.90.

That construction is what makes the index useful for comparing years. The dollar amount changes, the basket does not, so the ratio of two index readings isolates price change. It also explains the main limit: the basket is fixed within each year and refreshed annually, so the index tracks the cost of a typical urban household's purchases, not your own.

The CPI-U covers all urban consumers, roughly 93 percent of the US population according to the BLS Handbook of Methods, and the annual averages used here come from the historical CPI-U table. The narrower CPI-W, built from households that depend on wage and clerical income, covers about 30 percent and drives Social Security cost of living adjustments. Both use the same 1982-84 base, which is why their index values sit close together.

The four figures two index readings produce

Four figures come out of two index readings and one amount. The adjusted amount scales money forward or backward by the index ratio. The cumulative rate states the same move as a percentage. The annualized rate converts that percentage into a per year figure. The purchasing power loss states how much of the starting dollar's buying power disappeared.

Each one answers a different question, and using the wrong one is the most common error in CPI work. A raise that matches the cumulative rate over 35 years is not a raise at all, because the buying power only holds steady. A raise that matches the annualized rate for 35 years is a large real increase. Contracts, court awards and union agreements each specify which of the four they mean.

Worked example: $100 from 1990 to 2025

Take $100 spent in 1990 and carry it to 2025. The annual averages are 130.7 and 321.9, so 321.9 / 130.7 = 2.4629. Multiply the amount by that ratio and $100 becomes $246.29. The same ratio says prices rose by (321.9 - 130.7) / 130.7 = 1.4629, or 146.3 percent in cumulative terms.

Spread over 35 years, that cumulative figure becomes 2.61 percent a year. The arithmetic is 2.4629 ^ (1 / 35) - 1 = 0.0261, which is the compound rate that turns 1 into 2.4629 over 35 steps. It is lower than the cumulative number divided by 35, because compounding does part of the work each year.

The purchasing power loss uses the same two readings upside down. 1 - 130.7 / 321.9 = 0.594, so the 2025 dollar buys 59.4 percent less than the 1990 dollar. Put another way, $100 in 2025 buys what $40.60 bought in 1990.

Cumulative change against annualized change

Dividing the cumulative rate by the number of years gives an arithmetic average, and it overstates the annual rate whenever the total is large. The 146.3 percent rise over 35 years averages 4.18 percent if you divide, but the true compound rate is 2.61 percent. The gap is not rounding. It is the difference between simple and compound growth.

Use the geometric form for anything that compounds: rent escalators, salary projections, savings targets. Use the simple division only as a rough sanity check on a single year's move. If a contract says 3 percent a year, the index after 35 years should be 2.81 times the starting value, not 2.05 times.

Short windows flip the ranking. From 2020 to 2025 the index moved 24.4 percent cumulatively, which annualizes to 4.46 percent, a faster pace than any other window in the table below. Long windows smooth out shocks, so quote the window with the rate.

Purchasing power loss runs the other way

Purchasing power loss and cumulative inflation are not mirror images, and mixing them up makes an adjustment look wrong. A 146.3 percent rise in prices cuts the dollar's buying power by 59.4 percent, not by 146.3 percent. The two numbers describe the same event from opposite ends.

The reason is that the base for each calculation differs. Cumulative inflation divides the change by the starting index. Purchasing power divides the change by the ending index, because the question is what today's dollar can buy relative to before. One percentage point of extra inflation costs slightly less than one percent of buying power: at 1 percent inflation the loss is 0.99 percent.

This matters for wages. A salary that rose from $50,000 in 2015 to $67,911 in 2025 kept pace with the index exactly, since 50,000 x 1.35823 = 67,911. It gained nothing in real terms, even though the paycheck looks 36 percent larger.

CPI-U, CPI-W and the chained index

Three indexes cover consumer prices, and they do not agree. The CPI-U is the headline series and the one most contracts reference. The CPI-W covers wage earner households, runs a little lower in most years, and is the series used for Social Security. The chained CPI, or C-CPI-U, allows the basket to shift as relative prices change, so it rises more slowly than the CPI-U over long spans.

Pick the series the agreement names, not the one that produces a friendlier number. A lease that says "the Consumer Price Index for All Urban Consumers" means the CPI-U, and substituting the chained index would understate the increase by a few tenths of a point a year, which compounds.

Annual averages against monthly readings

The index is published as a monthly series and as an annual average. A contract that references a calendar year wants the annual average, which is the mean of the twelve monthly readings. That is the figure used everywhere on this page, and it is the one that avoids the noise of a single month.

Monthly readings matter for short adjustments. A rent increase tied to "the CPI published for the twelve months ending in September" uses a specific monthly release, not the annual average, and the two can differ by a few tenths of a point. Read the clause before you pull the number.

The table below lists annual averages at ten year intervals. Each row also shows what $100 from that year buys in 2025, which is the same ratio the calculator returns.

Where a CPI adjustment misleads

An index adjustment answers one question well: what did the same basket cost in two periods. It answers several other questions badly. It does not measure asset prices, so it will not tell you whether a house or a stock portfolio kept pace. The Federal Reserve targets inflation of 2 percent on a different gauge, the price index for personal consumption expenditures, which is a reminder that the 2 percent goal is not the CPI figure. It does not adjust for where you live, since the national average blends a Manhattan rent with a small town one.

The index also says nothing about quality change. A 2025 television at $600 is not the same product as a 1990 television at $600, and the BLS handles that with hedonic adjustment rather than a straight price comparison. For household budgeting the index is a rough anchor, not a forecast.

The practical rule: use the CPI to restate a past amount in current dollars, to check whether a raise or a rent increase kept pace, and to index a contract that names the series. Do not use it to project next year's prices, and do not use it as a discount rate for an investment.