- For example, see Czinkota and Ronkainen, (1997).
Quality change and new products
9.1 The quality of goods and services purchased by households can vary over time. Some products may improve in quality with new features or better services whereas some products may decrease in quality with smaller package sizes or a lower level of customer service. Quality adjustment is a conceptual requirement of a Consumer Price Index (CPI). The international CPI Manual (ILO, 2004) advises that failure to pay proper attention to quality changes can introduce serious biases into the CPI. Quality adjustment bias arises from the statistician's inability to perfectly account for changes in the quality of items over time.
9.2 While the quality adjustment issue raises important conceptual and practical challenges, the important role assigned by macroeconomic policy makers to price statistics underlines the priority that should be attached to ensuring that price statistics are not distorted by inappropriate quality adjustment procedures. The ABS has a number of strategies in place to minimise quality adjustment bias.
9.3 The objective of the CPI is to measure pure price change over time, so ideally identical goods and services should be priced from one period to the next. This is called pricing to constant quality. However, in practice, new products appear on the market frequently and replace older products. These new products have different attributes (or quality). For price index purposes, it is necessary to measure these changes in quality, and to remove any change in price attributable purely to the change in quality, from the inflationary movement in the price.
9.4 The concept of quality used in the Australian CPI is based on the notion of consumer utility. Quality change is measured by reference to the expected value to the consumer of the changes. Although it is not always possible to achieve this in practice, it is the principal guideline in making decisions about quality change.
9.5 The term quality embraces all those characteristics in a good or a service that a household values or from which it derives utility. Thus the problem is to identify those characteristics that households value, to make an estimate of the value of those characteristics, and to measure the change in those characteristics so that their effect can be removed when calculating price movements. When used in this context, quality encompasses all attributes of a product, including quantity.
9.6 Regardless of the difficulty in estimating the contribution of the change in quality to the change in the observed price, it must be clearly understood that some estimate has to be made either explicitly or implicitly. The following section describes the types of quality adjustments used by ABS pricing statisticians. A detailed description of the choice between quality adjustment methods is described in section 7.116 - 7.124 of the international CPI Manual (ILO, 2004).
Explicit quality adjustments
9.7 Explicit quality adjustments use information to apply a direct estimate of the effect of the quality change on price. These adjustments are generally considered to be more reliable, although they are more resource intensive in order to gather and apply all the relevant information. There are a number of methods of deriving explicit estimates. The suitability of explicit quality adjustments depends as much on the method used as on the availability of appropriate data to implement the methods. Explicit quality adjustments include quantity, option cost, expert judgement and hedonic modelling.
9.8 The quality change may take the form of a change in the physical characteristics of the product that can easily be quantified, such as change in weight, dimensions, purity, or chemical composition of a product. This quality change can be removed by applying a quantity adjustment.
9.9 To illustrate the process used to adjust for changes in the quality of items priced in the CPI, consider the case of a change in the size of a can of tomato soup. In this example, Acme brand tomato soup is priced in three periods (1, 2 and 3) and the size of the can is reduced from 440gms to 400gms between period 2 and period 3:
Using the observed prices produces the following measures of price change:
Percentage change from period 1 to period 2 = (2.78 - 3.09)/3.09 x 100 = -10.0%
Percentage change from period 2 to period 3 = (2.85 - 2.78)/2.78 x 100 = 2.5%
Percentage change from period 1 to period 3 = (2.85 - 3.09)/3.09 x 100 = -7.8%.
9.10 However, this does not provide a measure of 'pure price' change because the item priced in period 3 is not identical to the item priced in the previous periods. What is required for period 3 is the non-observed price of the period 2 can size (440gms) had it been available in period 3. This price can be estimated by adjusting the period 3 price by the ratio of the item's weight in period 2 to its weight in period 3, giving a quality adjusted price of $3.14 ($2.85 x 440/400).
Using this adjusted price in period 3 results in the following correct measures of price change:
Percentage change from period 1 to period 2 = (2.78 - 3.09)/3.09 x 100 = -10.0%
Percentage change from period 2 to period 3 = (3.14 - 2.78)/2.78 x 100 = 12.9%
Percentage change from period 1 to period 3 = (3.14 - 3.09)/3.09 x 100 = 1.6%.
9.11 After adjusting for the reduction in quality between periods 2 and 3, the rise in the observed price of 2.5% has been translated into a pure price increase of 12.9%. Similarly, the measure of price change between periods 1 and 3 has been changed from a fall of 7.8% to a rise of 1.6%.
9.12 Similar adjustment procedures can be used for other quality changes, the only issue being how to determine a suitable quality measure. For example, changes in the alcohol content of spirits could be allowed for, simply by adjusting the price proportionally for the change in the alcohol content. More difficult would be the handling of changes in the meat content of sausages or the salt content of margarine.
9.13 Of course, there are limits to the application of this approach. For example, it would be inappropriate to replace a medium sized can of tomato soup with a large or small sized can since price typically falls per unit of weight with significant increases in the container size. The samples should comprise relatively homogeneous items of similar sizes and quality.
9.14 The situation becomes more complicated with evolutionary goods such as high-technology commodities, motor vehicles and communication devices. Quality changes for these items are very hard to measure. A detailed explanation on the techniques used to measure quality changes for these items is set out in Price collection of this manual.
Differences in production or option costs
9.15 It is possible to try to measure the change in quality by the estimated change in the costs of producing the two qualities. The estimates can be made in consultation with the producers of the goods or services, if appropriate. This method, like the other quantity adjustments, will only be satisfactory when the changes are fairly simple such as the addition of some new feature, or option, to an automobile. Consider the addition of a Global Positioning System (GPS) navigation device to a new motor vehicle. In period 1 the price is obtained for the motor vehicle without the GPS option. In the next period GPS becomes standard for all motor vehicles. In order to accurately reflect the quality change, it is necessary to determine the cost of the option and the take-up rate by consumers. This information can be obtained from the dealer and/or manufacturer and used to determine the quality adjustment.
9.16 The use of expert views to estimate the value of the quality change is sometimes used for highly complex items where alternative methods are not feasible. Experts are guided with regard to the nature of the estimate required. This approach is generally not used in the CPI apart from estimates of quality change in motor vehicles where the option cost approach may not be feasible (i.e. the new feature was previously not an option). The expert judgement approach is generally applied using the Delphi method.¹ The Delphi method uses a panel of prices experts who each provide their estimate of the estimated value of the quality difference. The median is taken of these estimates and any estimate that is considered extreme is sent back to the expert concerned, who is asked to account for possible reasons behind the difference. The Delphi method is time-consuming and expensive, but it reflects the care needed in such matters.
Hedonic modelling technique
9.17 For more complex quality adjustment needs, statistical techniques such as hedonic modelling may be used. Hedonic modelling involves the use of a regression equation in which prices from an array of different varieties of a product are the dependent variables, and the price determining characteristics of that product are the independent (or explanatory) variables. The estimated parameters from the regression provide implicit prices for each of the price determining characteristics of the good. In simple terms, hedonic modelling divides a good or service into its component characteristics, and uses these characteristics as explanatory variables for the price.
9.18 Although this form of modelling allows for direct estimation of the amount of quality adjustment, large amounts of data and many calculations are required, which is expensive. An additional problem is that hedonic modelling techniques are not readily able to deal with quality changes that are not easily quantifiable, such as the handling characteristics of a car, the quality of medical care, or whether a variety of clothing is in or out of fashion.
Implicit quality adjustments
9.19 Implicit quality adjustments are easy to implement but care needs to be taken to ensure the quality adjustment is done appropriately. Rather than making a direct adjustment for quality described in the explicit adjustment above, an option is selected which automatically sets how the quality adjustment between the previous and current product is applied. These include directly comparable, not directly comparable and overlapping prices adjustments.
9.20 This option is used when the difference between the two products is assumed to be all price change and the quality difference is judged negligible. Directly comparable does not apply any quality adjustment and records any observed price difference between the two products as actual price change. This is most likely to be the case when the change in quality is described in terms of additions or deletions to the original product. One example could be a women's winter jacket. The price observed in the first period was for a wool blend jacket with four buttons and a large collar. In the second period this item was no longer available and replaced with a wool jacket with five buttons and a smaller collar. The utility to the consumer was judged to be the same and therefore the item was deemed directly comparable. Similarly a washing machine priced in the first period may be replaced with another with the same features but a different product code and slight styling change.
Not directly comparable
9.21 The purpose of this option is to link in the new product when the relative qualities of the original and replacement products cannot be compared in any meaningful way. In this case, an imputed price will be calculated for the old product for use in the current period's index calculation, and the price collected for the new product will be stored for use as the 'back price' in the following period. This type of quality adjustment is more common in commodities where the pace of innovation is high with continual changes in the characteristics of products such as televisions etc. Care is taken when applying this quality adjustment to ensure that the observed price difference between the original and replacement products is not due to one product being on sale, as this is not treated as quality change.
9.22 Where an item becomes permanently unavailable and needs to be replaced, a method is used to 'return the relatives' in order to not bias the index downwards. This means when an item goes on special and then is no longer available for purchase in subsequent periods, the price relative will be returned to its pre-special level before being replaced by another, non-comparable item, which will be priced into the future. This is to account for the fact that specials are temporary. If this adjustment was not made the price relative would be permanently biased downwards as a result of a not directly comparable item being chosen for replacement. This method is often used for goods that are rapidly changing and subject to heavy specialling activity such as clothing and furniture.
9.23 This option is used when both products are available on the market at the same time and the ratio of the prices of the new to the old quality should reflect their relative utilities to consumers.
9.24 Suppose a washing machine 'Model A' has been available on the market for a number of years but is now being replaced by a new model 'Model B'. In period 1 'Model A' washing machine retailed for $450. In period 2 'Model A' retails for $500 but the new model 'Model B' is available with extra features such as greater water efficiency for $530. In period 3 only 'Model B' is available for $550. If no quality adjustment was applied the price change from period 1 to period 3 would be overstated. The overlapping price quality adjustment uses the price change of 'Model A' from period 1 to period 2 and the price change of 'Model B' from period 2 to period 3. The difference in price between 'Model B' ($530) and 'Model A' ($500) of $30 is treated a quality change. The price change from period 1 to period 3 is $70. This is described in Table 9.1.
|Model||Period 1||Period 2||Period 3|
Using the observed prices produces the following measures of price change:
Price change from period 1 to period 2 = $500 - $450 = $50
Price change from period 2 to period 3 = $550 - $530 = $20
Total Price change from period 1 to period 3 = $50 + $20 = $70.
9.25 However, this method is not used very extensively because the requisite data are seldom available. Care is taken when applying this quality adjustment to ensure that the observed price difference between the original and replacement products is not due to one product being on sale, as this is not treated as quality change.
9.26 If there are no overlapping prices, or those prices are not normal, then quality adjustment becomes more difficult. It might be possible to use the last available price of the replaced item or to use estimates of differences in manufacturing costs. Again, using manufacturing costs will only be appropriate if costs broadly correlate with consumer utility.
Other issues related to quality change
9.27 There are other circumstances where the use of price differentials as indicators of quality differentials may not be appropriate. Examples include items that are heavily subsidised or regulated, such as public education and pharmaceuticals.
9.28 The quality of the service in which a product is delivered is important. Purchasing 2 litres of milk at the local convenience store is a different quality of service than at a supermarket, even if the product is exactly the same. Consumers substituting to different outlets will be treated as a quality change, not a price change. Price change is measured by matching products from the same type of outlet over time.
9.29 For some types of quality change, it is doubtful if any accurate measure of the change can be calculated. For example, in the case of services, consider changes in medical operating procedures (e.g. keyhole surgery) that involve less pain and a speedier recovery, or educational services making a greater use of computers. In these cases generally no quality adjustments are applied.
9.30 One important area of quality change is that arising from governmental regulations. It is ABS practice that, unless these changes clearly affect the level of household utility, they are not treated as quality changes. An example of this practice is that any higher price for motor vehicles occasioned by mandatory pollution requirements is regarded as a price increase, not a quality improvement.
9.31 An important issue is whether a change to an item should be regarded as a quality change to an existing item or the creation of a new item. The simpler approach is to assume that the item is new, and to include it into an existing price sample. However, this practice implicitly assumes that the difference in quality is equivalent to the price difference. Clearly, if it is assessed that a price differential is not a reliable indicator of quality or household utility differentials, then some other appropriate quality adjustment should be made.
New goods and services
9.32 From time to time, major changes in existing products and services take place, or new products and services become available on the retail market and begin to account for a significant share of household expenditure. Some examples in recent years are tablets, 3D televisions, smart phones and electronic books. In these cases, careful consideration is given to whether these new goods or services should be priced for the CPI.
9.33 If a new product or service is deemed to be a completely different category of product (i.e. a new expenditure class) from any of the goods and services already included in the CPI, its inclusion would be considered only during one of the periodic reviews of the index where updated weighting patterns at the published level were available. The inclusion of television sets in the 1960s is a good example of this. However, where a new product or service falls within the definition of an existing expenditure class (e.g. the introduction of colour television sets, or mobile telephones), the issue is when and how to start measuring these price movements for the CPI. Normally, the decision is made after considering the following factors.
- The product's share of the market. This has to be substantial before there is a reason to introduce a new item;
- Whether the product is firmly established, and expected to become a permanently significant item of expenditure, or is merely enjoying high sales temporarily because of novelty value; and
- Whether a normal price structure has been established, that is a price structure that is not unduly influenced by factors such as prestige, novelty value, or scarcity of the product.
9.34 In general, a conservative approach is taken when dealing with the introduction of new goods and services into the CPI. They are introduced into existing expenditure classes only after it is deemed that they have become widely available to the buying public, have become a permanent part of household expenditure, and their price structures are free from premiums attributable to novelty value or scarcity. All introductions of new items are handled by including the new item into the index so that its introduction does not affect the level of the index.