Quality change in price indices is the problem of telling a genuine change in price apart from a change in what the buyer gets. When a new phone model with a better camera replaces an old one at a higher price, part of the increase pays for the improvement and only the rest is inflation. Statistical agencies such as the Bureau of Labor Statistics, the Office for National Statistics and Eurostat devote much of their effort on the consumer price index and the producer price index to this problem.
The members of this family form a toolkit. The matched-model method is the starting point: price exactly the same item over time. When an item changes or disappears, quality adjustment is needed, either explicitly, by putting a value on the difference, or implicitly, by inferring it from other prices. The new goods problem arises when products have no predecessor at all. Owner-occupied housing is a special case where the difficulty is less about quality than about what the price of housing services should be.
Matched-Model Method
The matched-model method prices exactly the same product specification, in the same outlet, in successive periods, so that measured price change is not mixed up with quality change. Price collectors record details such as brand, model, size and features, and the index compares like with like. This is the foundation of almost every consumer and producer price index.
The method works well for stable products but runs into trouble where products turn over quickly. When an item leaves the market, it must be replaced, and the comparison is no longer matched. If only surviving items are tracked, the sample becomes unrepresentative: older models often fall in price before they disappear, while new models enter at higher prices with better features. Ignoring these entries and exits can bias the index in either direction. Fast-changing goods such as electronics and clothing are where the problem is greatest, which is why quality adjustment methods are needed.
Quality Adjustment
Quality adjustment is the procedure for removing the effect of a change in quality when a product in a price index sample is replaced by a different one. The goal is to record only the pure price change, the change in price for a constant quality, so that the index measures inflation rather than improvements or deterioration in products.
Methods fall into two groups. Explicit methods estimate the value of the quality difference directly and adjust the price of the old or new item accordingly. Implicit methods avoid valuing the difference and instead infer it from price movements of other items or from overlapping prices. The choice matters: if an improved product's higher price is all counted as inflation, the index overstates price growth; if a decline in quality, such as a smaller pack sold at the same price, is missed, the index understates it. The hedonic adjustment method is the best-known explicit technique, and international manuals for CPIs and PPIs describe the full range.
Explicit Quality Adjustment
Explicit quality adjustment means directly estimating the value of a quality difference between an old and a replacement product, then adjusting the price comparison by that amount. Several techniques are used.
- Quantity adjustment. Where only the size changes, as with a smaller packet, prices are compared per unit.
- Option costs. Where a feature once sold as an extra becomes standard, its price is taken as the value of the improvement.
- Hedonic regression. Prices of many models are related statistically to their characteristics, so the value of each feature can be estimated.
A common hedonic specification is:
where is the price of model , its characteristics and the estimated value of each. If an old laptop cost 1,000 and its replacement costs 1,100 but has extra features worth an estimated 60, the quality-adjusted price change is per cent rather than 10 per cent.
Implicit Quality Adjustment
Implicit quality adjustment infers a quality change indirectly instead of valuing it. It is used when there is no information to put a value on the difference between old and new items.
In the overlap method, when old and new items are on sale at the same time, the price ratio between them in the overlap period is treated as the quality difference. In class mean imputation, the price change for a replaced item is assumed to equal the average change for comparable items with no quality change, sometimes limited to other replacements. The link-to-show-no-change approach simply assumes the replacement involves no price change, attributing the entire price difference to quality, which can understate inflation. Implicit methods are cheap and widely used but rest on assumptions that may not hold: overlap prices may reflect old models being cleared at a discount, and price changes often cluster when new models launch.
New Goods Problem
The new goods problem is the difficulty price indices have in capturing the benefit of products that did not exist before. A fixed basket cannot include a product that was not available in the base period, and when a new good is introduced, its arrival is often a gain to consumers that a price index does not record. Before it existed, its effective price was in a sense infinitely high.
Agencies usually add new goods when the basket of goods is updated, often years after launch, by which time much of the early price fall has happened. Mobile phones, for example, entered some CPIs long after they were in wide use. Delayed entry and the missed welfare gain tend to make a cost-of-living index overstate inflation; the 1996 Boskin Commission put the total upward bias in the US CPI at about 1.1 percentage points a year. Responses include more frequent basket updates, larger samples that pick up new products sooner, the use of scanner and web-scraped data, and research methods that estimate a virtual price before launch.
Owner-Occupied Housing
Owner-occupied housing is the cost of living in a home one owns. It is a large part of household spending, but owners do not pay rent, so a price index must choose how, or whether, to include it. Three approaches are used.
- Rental equivalence. Owners are treated as renting the home to themselves, valued at market rents for similar properties, an imputed rent.
- User cost. The cost is built from mortgage interest, depreciation, taxes and expected price gains.
- Net acquisitions. Only purchases of new dwellings by households are included, usually excluding land.
The UK's CPIH and the US CPI use rental equivalence; Australia's CPI uses net acquisitions. The euro area's HICP excludes owner-occupied housing, and the ECB's 2021 strategy review recommended adding it over time using the net acquisitions approach.
External links
Ref 1100001830