Consumer Price Index

Construction begins with expenditure surveys that establish what a reference household actually buys and in what proportion. Those expenditure shares become the weights, and the resulting basket is priced repeatedly at retail outlets, service providers and rental markets. The index level is the cost of buying that basket at current prices divided by the cost of the same basket at base-period prices, multiplied by 100. Inflation is then the percentage change in the index between two dates: the later index level divided by the earlier one, minus one.

Because quantities are held at their base-period levels, the standard CPI is a Laspeyres index. Weights are refreshed only periodically, so between revisions the basket reflects consumption patterns that are already somewhat dated. Sub-indices for food, energy, shelter, medical care and transportation aggregate up to the headline figure, which is why analysts can decompose a monthly print into contributions by component.

Four well-documented distortions push a fixed-basket index above the true change in the cost of maintaining a given standard of living. Substitution bias arises because households shift toward relatively cheaper goods, while the basket assumes they do not. Quality bias appears when a product improves without a proportionate price increase, so part of the measured price rise actually buys more utility. New-product bias reflects the lag before innovations enter the basket, typically after their steepest price declines. Outlet bias arises when shoppers migrate to discounters that surveys underweight.

Statistical agencies respond with hedonic regressions that strip out quality change, more frequent weight updates, and chained indices that let quantities adjust between periods. A Paasche index, which uses current-period quantities, tends to understate inflation for the mirror-image reason; a Fisher index, the geometric mean of the two, sits between them.

Published CPI series drive real money. Inflation-linked bonds such as US TIPS index principal and coupons to a lagged CPI reading, so the breakeven spread between nominal Treasuries and linkers is a market-implied CPI forecast. Social security payments, union wage agreements, commercial lease escalators and tax bracket thresholds are frequently tied to a specific index variant, which makes the choice between headline, chained and core versions fiscally material.

CPI also converts nominal magnitudes into real ones: deflating nominal wages, revenues or interest rates by the index reveals purchasing power. Note that CPI is not the only inflation gauge. The GDP deflator covers all domestically produced output with current-period weights and excludes imports, while producer price indices measure prices at earlier stages of production. The Federal Reserve's numerical objective is stated in terms of PCE inflation rather than CPI.

Expect to price a small basket in two periods, build the index, and convert the change into an annualised inflation rate. Expect also to work backwards: given nominal growth and an inflation figure, isolate real growth, or use the relation among nominal rate, real rate and expected inflation.

Beyond arithmetic, candidates should be able to state the direction of bias for Laspeyres, Paasche and Fisher constructions and name the mechanism behind each, explain why headline readings are volatile relative to trimmed-mean, median or core measures, and identify whether a given inflation episode looks demand-pull or cost-push from the behaviour of components. Recognising that different countries publish different variants, and that the choice of index affects indexed payments and measured real returns, is the practical takeaway.