What CPI is, in plain terms
CPI stands for Consumer Price Index. It is a statistic that summarizes how prices for a selected “basket” of goods and services change over time. The core idea is simple: compare the cost of the basket in a later period with the cost in an earlier period, then convert that change into an index level (and often into a percentage change).
A “worked example” is useful because it shows exactly what inputs are needed (basket items, prices in two periods, and how strongly each item matters) and how the calculations flow from those inputs. Different CPI methodologies can lead to different results, even when the underlying prices look similar.
How the CPI calculation works (worked example setup)
To keep the example self-contained, we will use a simplified CPI method:
- Basket: a few items only (real CPI uses many items).
- Weights: each item has a weight representing its relative importance in the basket.
- Two time points: a baseline period (Period 0) and a comparison period (Period 1).
- Price relatives: compute each item’s price change from Period 0 to Period 1.
- Weighted average: combine the item changes using the basket weights.
- Index conversion: if the baseline CPI is set to 100, then the Period 1 CPI is 100 × (1 + weighted percent change).
Assumptions (state them up front):
- Weights stay constant across the two periods.
- Each item’s “price” is a single representative price per period.
- We ignore taxes, rebates, and regional differences.
- We treat quality and product changes as identical across periods.
Worked numerical example of CPI
Assume a simplified CPI basket with two items:
- Item A: weight = 0.60
- Item B: weight = 0.40
- Weights sum to 1.00
- Baseline CPI (Period 0) = 100
Assume prices are:
- Item A: Period 0 price = 50, Period 1 price = 55
- Item B: Period 0 price = 100, Period 1 price = 103
Step 1: Compute each item’s percent price change.
- Item A percent change = (55 − 50) / 50 = 0.10 = 10%
- Item B percent change = (103 − 100) / 100 = 0.03 = 3%
Step 2: Compute the weighted average percent change.
- Weighted percent change = (0.60 × 10%) + (0.40 × 3%)
- = 6% + 1.2% = 7.2%
Step 3: Convert to an index value.
- Period 1 CPI = 100 × (1 + 7.2%)
- = 100 × 1.072 = 107.2
So, in this worked scenario, CPI rises from 100 to 107.2, which corresponds to a 7.2% weighted increase in the assumed basket prices.
Material limitations and failure modes
Even if the arithmetic is correct, CPI can still misrepresent what people experience. Common limitations include:
-
Substitution effects (basket rigidity): If one item becomes more expensive, people may buy substitutes. A CPI that keeps weights fixed can overstate cost-of-living changes.
-
Quality changes and versioning: If an item improves (or worsens) in quality, “price change” may not correspond to “the same thing getting more expensive.” Adjustments for quality are method-dependent.
-
Representative price problems: Using a single price per period per item can be misleading if prices vary widely by location, seller, or timing.
-
Index construction choices: CPI uses specific mathematical structures for aggregating price changes and sometimes for handling items that enter/exit the basket. Different choices can shift the result.
-
Data and measurement errors: Missing items, inconsistent measurement, or inconsistent time windows can introduce error.
These limitations are a key failure mode: the CPI number is not just “an average of prices,” but a result of explicit methodological decisions.
How to independently verify the CPI logic you are using
A practical verification approach is to check that your own calculation follows the same principles you assume:
- Confirm the basket definition: which items are included and how they are weighted.
- Confirm the baseline and comparison periods.
- Confirm the price series definition: what “price” means for each item.
- Recompute the index steps: price relatives, weighted average, then index conversion.
- Stress-test assumptions: try alternative weights or plausible price definitions to see how sensitive the final CPI is.
If two sources produce different CPI numbers, you can usually trace the discrepancy to differences in basket composition, weights, price collection, or methodology—not just to “market reality.”