Price indices are designed to measure “pure” price changes. This is accomplished by keeping quantities fixed, allowing prices of comparable products only to vary over time. Any change in value due to a quality change should be counted as a volume change, since improved quality etc. provides users with additional utility or satisfaction.
Quality adjustment
Frequently, old varieties disappear and new varieties are introduced on the market. Effort must be made in order to determine whether new varieties differ from the old model in any aspects related to quality – which is to be treated as a change in volume, and not in price.
E.g., a new laptop (model) is introduced on the market. Compared to last year’s model the price is 10 percent lower. However, the memory is bigger, it is faster, and the weight is less – the standard is higher. Should the price index reflect a 10 percent decrease in prices on computers? The answer is no, the price is falling by more than 10 percent if the improved quality is taken into account.
The statisticians are facing the problem of how to adjust for quality changes in order for the price index to reflect ‘pure’ price change.
Several techniques of quality adjustments can be applied in price index computations.
(i) One approach is to make direct estimate of the value of the quality effects and then remove this difference from the price change.
(ii) Another approach is to use statistical regression models (hedonic models) to estimate the value of the quality change. In the models the prices paid for the products are regressed on the various characteristics to estimate the contribution to the final price of each characteristic. When characteristics change, the value of the quality change can be estimated from the regression equation.
(iii) Statistical methods for estimating missing values can also be applied to make adjustments for quality changes. If the old and the new model are available in the market at the same point of time, the observed difference in their market prices can be used as an estimate of the value of the quality difference. The old variety’s price change will then be used as the representative item in the index throughout the ‘overlap period’, while the new variety’s price change will be used in the next. It turns a little trickier if the if the old variety is no longer available when the new one is introduced – no ‘overlap period’. A price change for the old variety is then estimated using the change for similar products, and for subsequent periods are the new variety’s price change used.
Even though extremely difficult to estimate the value of quality changes, it should be noted that it is not obvious that the effect of quality changes is underestimated. It is not obvious that we are overestimating price changes. Some of the adjustment methods used might actually overestimate quality changes, resulting in underestimation of price changes.
Substitution bias
In addition to the problem of changes in quality and user satisfaction, a number of other potential biases occur in fixed-base price indices. One is the substitution bias that occurs as a result of keeping the quantity fixed in the Laspeyres and other fixed-base price indices. Such indices do not allow for product substitution taking place when relative prices changes. Shift in purchasing patters are not reflected, and become more serious as time passes by. Furthermore, substitution occurs across outlets, while price surveys usually have fixed outlet samples. When new outlets that offer lower prices open, purchasers tend to shift to the new outlets, but this will not be reflected in the price surveys. In fact, no comparisons are made of the prices between these new outlets and the old ones.
New good bias
Furthermore, a ‘new goods bias’ occurs when new goods not are introduced into the indices in a timely fashion. Many new products with a higher consumer satisfaction are introduced into the market and proceed to decline in price. Such products are usually not brought into the samples of goods and services used for the estimation of the price indices until after they are established in the market and their price levels have stabilized (which may take several years). |
RSS