Hedonic Quality Adjustment Methods For Microwave Ovens In the U.S. CPI

Paul R. Liegey⁽¹⁾

Preface

The U.S. Consumer Price Index (CPI) for Major Appliances would have increased 0.6 percent instead of the official index increase of 0.8 percent if hedonic quality adjustment methods had been applied to microwave oven (substitution) price changes from August 1999 through April 2000 (see attachment 1). This quality adjusted empirical result provides a more refined measure of price inflation for Major Appliances since there are a greater number of "pure" price changes used to calculate this item stratums' price index.

Of the three different approaches or methods that use the results from hedonic regression models to quality adjust price indexes, BLS employs the 'matched model' method in its official indexes.² This method controls for quality changes based on the difference in product specifications or characteristics between two items when a substitute observation, or quote, occurs in the price index sample. It is important to note that under the 'matched model' approach only substitution price changes, or quotes, are eligible for hedonic quality adjustments.

As previously announced, the Bureau of Labor Statistics (BLS) is extending the use of quality adjustments derived from hedonic models in the CPI. A hedonic model decomposes the price of a consumer product into implicit prices for each of its important features and components, thereby providing an estimate of the value for each price influencing feature and component.

Effective with the CPI for July 2000, BLS has extended hedonic quality adjustments to Microwave ovens and Refrigerators, two products in the Major Appliances item stratum. The relative importance (share of weight), as of December 1999, for this stratum was 0.205 percent in the CPI for all Urban Consumers (CPI-U) and 0.236 percent in the CPI for Urban Wage Earners and Clerical Workers (CPI- W). Within Major Appliances, Microwave ovens are estimated to represent 11 percent of the weight and Refrigerators about 33 percent. In October 2000, BLS will further extend hedonic quality adjustments to Washers and Dryers, two other products in the Major Appliances item stratum. The remaining products in this stratum—those that will not be subject to hedonic quality adjustment at this time—include Freezers and Stoves and Ovens.

Background

In 1947, Raytheon demonstrated the world's first microwave oven and called it a "Radarange."³ The first microwave ovens cost between two and three thousand dollars and were sold primarily to the commercial marketplace. By the early 1950s, U.S. appliance makers began showing interest in the microwave oven since it had the ability to provide more flexibility to the consumer by reducing the time needed to cook different meals. In 1955, the Tappan Stove Company—under a licensing agreement with Raytheon—brought the first consumer microwave ovens to the U.S. market with a price tag of $1,300. These first consumer microwave ovens were more compact and less powerful than their "Radarange" predecessors and reflected only modest sales at their price range.

In 1965, Raytheon acquired Amana Refrigerators, Incorporated—an appliance maker with a well established distribution channel—and, in 1967, introduced to the U.S. market the first "countertop" model of microwave ovens that sold for $495 retail and were smaller, safer and more reliable than previous models. By 1986, 25 percent of U.S. households owned a microwave oven, up from less than one percent in 1971.⁴ Assuming microwave oven penetration into U.S. households was constant during this 15 year period, about 12 to 13 percent of U.S. households would have owned a microwave oven in 1978.

The microwave oven was introduced into the CPI sample in 1978.⁵ Since its' inception to the CPI, microwave ovens have been represented in three item stratum price indexes over three CPI revisions:

• With the release of the revised CPI in January 1978, microwave ovens became part of the Stoves, dishwashers, vacuums, and sewing machines item stratum price index. The index level for this item stratum in December 1977 was 100.0 (its base year), and increased to 117.4 by December 1986 — an overall price index increase of 17.4 percent for the 9 year period.

• With the release of the revised CPI in January 1987, microwave ovens became part of the Stoves, ovens, dishwashers, and air conditioners item stratum price index. The index level for this item stratum in December 1986 was 100.0 (its base year), and decreased to 91.6 by December 1997 — an overall price index decline of 8.4 percent for the 11 year period.

• Finally, with the release of the revised CPI in January 1998, microwave ovens became part of the Major Appliances item stratum price index. The index level for this item stratum in December 1997 was 100.0 (its base year), and through March 2000 is 97.9 — an overall price index decline of 2.1 percent for the most recent two and a quarter year period.

Today, more than 90 percent of U.S. households own a microwave oven.⁶ As consumer expenditures on microwave ovens have increased through time in the U.S. economy, so too have their representation in the CPI — currently accounting for 11 percent of the weight in the Major Appliances item stratum price index.

Microwave ovens were selected as a product that would benefit from hedonic regression modeling since manufacturers provide a vast range of sizes, configurations, and features. In the Consumer Digest’s 2000 Annual Buying Guide, the latest annual review on microwave ovens notes that

These newer models of microwave ovens are included for price index calculations only if older models being used in the price index sample are no longer available (in the CPI outlets) for CPI pricing. When a newer model replaces an older model in the CPI, the price change that is used in the index is referred to as substitution price change. Substitution price change can be either "pure" (directly compared or quality adjusted) or "imputed" (non-comparable).

From August 1999 through April 2000 Major Appliances employed, on average, 215 price changes to calculate its U.S. level monthly price index (see attachment 2). Approximately 10 percent, or 22, of these price changes were substitution price changes. In comparison, microwave ovens accounted for, on average, 39 of the 215 price changes used to calculate Major Appliances index, and averaged (just) four substitutions per month from August 1999 through April 2000.

Data

Sample selection for the Hedonic Model

The official CPI sample of microwave oven prices used to calculate the Major Appliances CPI was too small for hedonic regression estimation. Using a process that mimics the official CPI sample selection process, an additional sample of 195 consumer businesses, or outlets, was chosen to augment the official CPI sample for microwave ovens. This additional outlet sample was used to select a sample of microwave oven prices. The supplementary sample was used only for estimating the hedonic regression model for microwave ovens.⁸

CPI field economists were instructed to collect a total of 390 microwave oven prices in the sample of 195 additional outlets. Individual microwave oven brand and models were selected by grouping all microwave ovens in a particular outlet into two groups—the "standard" microwave ovens and "better model" or higher quality microwave ovens. Once categorized into these two groups, the CPI field economist was instructed to select a "good selling" microwave oven from each of the "standard" and "better model" groups.

About 73 percent of the additional sample price quotes that were collected for microwave ovens had price and characteristic data that could be used in the regression model. The most common reason that CPI field economists could not collect the additional sample price quotes for microwave ovens was respondent refusal.⁹ A total of 381 prices—and characteristic descriptions—were used to estimate the hedonic model for microwave ovens. This total sample consists of 98 official CPI observations and 283 additional sample observations.

Price and Characteristic Data for the Hedonic Model

All of the price and characteristic data used for the microwave oven hedonic model was captured on CPI data collection documents, or checklists, for this item (see last attachment).

The prices that were collected for the microwave oven sample represent "retail offer" prices. As the name suggests, a retail offer price represents what a consumer business is willing to sell an item for which may, or may not, differ from the transaction price—what a consumer actually paid for the item. Retail offer prices, like transaction prices, may change through time depending on whether the item being sold is offered at a "regular" price or a "sale" price.

The set of quality attributes collected for each of the 381 microwave ovens in the sample are represented on the CPI checklist for this item (see last attachment). In each of the quality characteristic categories, CPI field economists selected the specific characteristic element that best described the item they were pricing. For example, if an Amana countertop model with a (oven) capacity of 1.6 cubic feet and cooking power of 1000 watts had been selected by the field economist, this would be designated on the CPI data collection document by selecting the A1, B1, D3, and E2 specification elements—see last attachment.

When possible, secondary source information such as manufacturer websites and consumer information magazines—including Consumer Digest’s and Consumer Reports—were used to verify the accuracy of the characteristic data collected on the CPI checklist for microwave ovens. Overall, the CPI field economists were able to provide complete, consistent, and accurate descriptions for most microwave oven observations included in the hedonic regression sample.

Model

The hedonic model that was specified for microwave ovens in this study resembles the categories of quality characteristics that are presented on the CPI checklist for this item (see last attachment):

ln P = b0 + b1 Type + b2 Capacity + b3 Brand + b4 Oven Controls + b5 Features + b6 Control + e.

The bi’s in this model represent the effect of the characteristics on the natural logarithm of price. The CPI prices that were collected in this sample represent "retail offer" prices, and approximately 37 percent of these prices were collected "on sale." The mean price for all microwave ovens in the sample was $207.75. The mean price for "regular" priced microwave ovens in the sample is $214.77 and the mean price for "sale" priced microwave ovens is $195.61. Since type of price (that is, regular or sale) is thought to have an impact on the overall price level, a dummy variable for sale price was included in the model to capture this effect, and its expected coefficient sign was negative.

A priori

Microwave Oven Types

Basically, two types of microwave ovens are sold in today's appliance market. Countertop models are more frequently advertised by manufacturers and retailers and accounted for approximately 76 percent of the microwave ovens in the sample. Consumers who have smaller kitchens and lack counter space can opt for an over-the-range model—this type of microwave oven accounted for the remaining 24 percent of the microwave oven sample. All over-the-range models come with an installation kit for mounting. In addition, since over-the-range models take the place of a range or stove exhaust hood, all come with a ventilation fan that exhaust both the range and microwave oven cooking gases. The mean prices for the 288 countertop models and 93 over-the-range models in the hedonic sample were $140.64 and $416.31, respectively. By way of comparison, a December 1999 Consumer Reports study on microwave ovens found that the countertop ovens they tested sold for $200 or less while over-the-range models typically sold for about $350 to $450.¹⁰ Dummy or indicator variables were created for both types of microwave ovens. The over-the-range variable was included in preliminary models and was expected to have a large, positive impact on price.

Both the size and cooking power characteristics of microwave ovens are also assumed to have a significant impact on microwave oven prices. Manufacturers and retailers advertise size in terms of cubic feet and cooking power in terms of maximum watts. Both size and cooking power assume a wide range of numeric values—see the CPI checklist for microwave ovens (last attachment). To accommodate this wide range of values, continuous variables were created for both cavity cubic feet and maximum watt cooking. It was assumed that the price of the microwave oven would increase with increasing values for one or both of these variables. For example, a 2.0 cubic foot oven with 1100 watts of cooking power is assumed to sell for a higher price than a 0.8 cubic foot oven with 700 watts of cooking power ceteris paribus.

A preliminary model was specified with two dummy variables, sale price and above stove, and two continuous variables, cavity cubic feet and maximum wattage (see attachment 3, Iterative regression 1). This model specification proved to explain a significant portion of the variation in (the natural log of) price with an R² of slightly more than 82 percent. While the magnitude, direction and significance of the parameter estimates in the preliminary model generally conformed with a priori expectations, somewhat low tolerance values for the cavity cubic feet and maximum wattage variables indicated that multicollinearity might be present in the model. Further investigation revealed that the pearson correlation coefficient, or measure of collinearity, for these two variables was positive and strong at 0.61—they tend to move together and can serve as a proxy for each other. The existence of multicollinearity causes the standard errors of the correlated variables to increase and the associated parameter estimates to be imprecise.¹¹ Subsequent variations of this preliminary model—not included in Attachment 3—revealed that the variable for cavity cubic feet provided a better overall fit for the model and was included in the final model.

Microwave Oven Brands

Brand

As an established leader in the production and distribution of microwave ovens, Amana was thought to have a positive impact on price. Inclusion of the brand variables with the previous model specification—see attachment 3, Iterative regression 2—yielded a slightly better fit with an R² of 87 percent. In the absence of strong a priori about brands, low tolerance values and statistically insignificant parameter estimates led to the exclusion of some of the brand name variables. Subsequent variations of this preliminary model—not included in Attachment 3—and inclusion of other quality category variables on the CPI checklist led to the following brands being included in the final model: Amana, Kitchenaid, General Electric, Whirlpool, Samsung, Admiral, Emerson, and Sanyo.

Interestingly enough, the recent report on the CPI by The Conference Board, notes that "In existing products, probably the biggest problem in quality change that the CPI never recognizes (is) the subtle, often gradual changes that are difficult to detect and to measure. These would include for example,…reduced incidence of repairs to household appliances…"¹² Coincidentally, a December 1999 Consumer Reports study on microwave ovens supplies data about percent of "repairs and serious problems" for microwave oven brands (see Attachment 4). This data was compared to the implicit price values estimated in the final hedonic regression model for microwave ovens to see if consumer valuation of brand was consistent across studies. Of the eight brands included in the final hedonic model and the 12 brands included in the Consumer Reports study, six brands were common to both studies. The following is an ordinal ranking of brand value for both studies results

• 1 = least repairs
• 6 = most repairs
• A = most positive contribution to price
• F = most negative contribution to price

A cursory comparison of the two studies reveals that brands in the hedonic model with the most positive implicit price values have more "repairs and serious problems" while brands with the most negative implicit price values have less "repairs and serious problems." This (limited) comparison of brand quality seems to suggest that repair histories do not (in and of themselves) influence relative price levels for microwave oven brands.

Microwave Oven Controls

Oven Control

In the "Type of Controls" specification category, dummy variables were created for electronic controls and manual controls. While most of the microwave ovens in today's appliance market possess touch sensitive or electronic controls, some manufacturers still offer models with push button or rotary dial manual controls. Almost 95 percent of the microwave oven sample contained microwave ovens with electronic controls, but a variable for manual controls was included in a preliminary model—see attachment 3, iterative regression 3—since a priori expectations were that only inferior quality models possessed this type of control. As expected, the parameter estimate for the manual controls variable was negative, remained robust in subsequent regressions, and was included in the final model.

Most microwave ovens can control the (total) cooking power or wattage—most commonly this feature is given in 10 percent increments ranging from a low power of 10 percent to a high power of 100 percent. The December 1999 Consumer Reports study on microwave ovens notes that "most models" have 10 power settings.¹³ Similarly, most of the observations in the microwave oven sample were described as possessing 10 power settings. Other values collected for this specification element were "manual setting," "not specified," and "not applicable." Since 10 power settings was the only specification element listed in the "Number of Power Settings" category and no other meaningful variables could be created, the 10 power settings variable was not included in the final model.

A less advertised characteristic on microwave ovens is the "Number of Pre-Set Programming Stages." This oven control allows the user to pre-set or pre-program the microwave oven to cook food at a certain power setting for a certain length of time and then automatically switch to a different power setting for another length of time. An example of this feature might be programming the microwave oven to cook a food for two minutes at defrost, three minutes at a high temperature and one-minute at a low temperature.

It was assumed that microwave ovens with a higher number of sequential cooking stages offered more convenience and greater value to the consumer than microwave ovens with a lesser number of sequential cooking stages ceteris paribus. However, the December 1999 Consumer Reports study on microwave ovens notes that "most models can be programmed to cook for at least two sequential stages and up to about 100 minutes in all."¹⁴

Dummy variables were created similar to the specification elements on the CPI checklist (see last attachment) for the "Number of Pre-Set Programming Stages." In particular, variables for 4 stage, 3 stage, and 2 stage programming were tested in a preliminary model—see attachment 3, iterative regression 3. While the parameter estimates for 3 stage and 2 stage programming were marginally negative and statistically insignificant, the parameter estimate for 4 stage programming was positive and statistically significant. Subsequent variations of this preliminary model—not included in Attachment 3—and inclusion of other quality category variables led to the 4 stage programming variable remaining in the final model.

Microwave Oven Features

Manufacturers and retailers of microwave ovens offer and advertise a variety of features on their products. Microwave oven features are represented in the "Cavity Features," "Cooking Features," and "Programming Features" specification categories on the CPI checklist (see last attachment) for microwave ovens. Dummy or indicator variables were created for all characteristics in these specification categories. Each category of variables was tested to see whether the overall fit of the model could be enhanced. Basic preliminary models used to test each category of features are presented in Attachment 3, iterative regressions 4 through 6.

Of the 19 dummy variables that were created for the three categories of feature specifications, only four of these variables—built-in sensor, auto sensor, convection cook, and convection broil—were used in the final regression model for microwave ovens. In addition, since these characteristics were not clearly delineated on the CPI checklist—or in the product descriptions at the retail outlets—the built-in sensor and auto sensor variables were combined into one variable called sensor and the convection cook and convection broil variables were combined into one variable called convection.

Sensor

Microwave ovens with convection cooking rely on a fan that gently circulates heated air over and around the food, producing golden brown exteriors and moist interiors. Convection microwaves ovens can function as normal microwaves ovens, as a regular convection oven, or a combination of the two at once. This feature was reported for about five percent of the microwave oven sample and was expected a priori to have a strong positive influence on price. Since the parameter estimate for convection was positive, statistically significant, and remained robust in subsequent variations of preliminary models—included and not included in Attachment 3—the convection variable remained in the final model.

Most other microwave oven feature variables were excluded from the final regression model because of their poor performance in preliminary and subsequent regression models. For example, all of the feature parameter estimates in the "Programming Features" specification category consistently reflected small nominal impacts on price and were statistically insignificant. These kinds of features are frequently touted by retailers and manufacturers as value enhancing but typically do not contribute (substantially) to the price—or value—of microwave ovens that possess them.

Microwave Oven Miscellaneous Specifications

Other CPI checklist specification categories that were examined for hedonic model development included optional installation kits for countertop models as well as delivery charges. These CPI categories were found to be, for the most part, not representative of the microwave oven market. On the few occasions that either the cost of optional installation kits or delivery charges were reported with the price of the microwave oven, these additional charges were "backed out" of the overall price so that just the price of the microwave oven was reflected in the sample data.

Similarly, the manufacturer warranty specification category on the CPI checklist could not be utilized in specifying the hedonic model for microwave ovens. Almost all microwave ovens that are sold in today's market come with a standard one year parts and labor manufacturer warranty. Since this was the predominant type of manufacturer warranty captured in the microwave oven sample, it could not be used in the final regression model.

Finally, the country of origin specification category on the CPI checklist was tested to see if the hedonic model for microwave ovens could be fitted with any of these variables. Country of origin represents the country in which the product—in this case microwave ovens—was constructed and is believed to serve as a proxy for the quality of a good and service. In the sample used in this study, over 90 percent of the 381 microwave ovens were constructed in six different countries. In particular, microwave ovens constructed in the USA, Japan, Korea, China, Singapore, and Thailand accounted for 42, 8, 16, 6, 10, and 9 percent of the sample data, respectively.

In the absence of a priori about country of origin, low tolerance values and statistically insignificant parameter estimates led to the exclusion of most of the country of origin variables. Only the parameter estimate for Korea was statistically significant and remained in the final regression model.

Control Variables

Various control variables were tested representing region of the country, city size, and type of business as defined in the CPI. The most consistently performing and statistically significant control variables that were included in the final microwave oven model were appliance stores, warehouse outlets, and A size cities(populations 1,000,000 or greater) in the Northeast andWest regions. The purpose of control variables are to minimize the unexplained variation that might remain after the model has been fitted with price determining characteristics.

Attachment 3, iterative regressions 1 through 7, are included in this study to give the reader a sense of how the hedonic model for microwave ovens progressed as more categories of CPI variables were considered for model inclusion. Iterative regressions—both included and not included in Attachment 3—were performed until the remaining parameter estimates in the model below exhibited relative robustness to the inclusion and deletion of other variables not included. The direction and magnitude of the parameter estimates seem reasonable, and the statistics pertaining to fit, significance, and collinearity are within generally accepted limits.

Hedonic Quality Adjustments and Price Index Simulation

BLS employs the 'matched model' method of quality adjustment in its official indexes. This method controls for quality changes based on the difference in product specifications or characteristics between two items when a substitute observation, or quote, occurs in the price index sample. It is important to note that under the 'matched model' approach only substitution price changes, or quotes, are eligible for hedonic quality adjustments.

During the study period from August 1999 through April 2000, a total of 39 microwave oven substitution price changes—an average of (just) four per month—occurred in the Major Appliances price index sample. For each microwave oven substitution price quote, differences in the specification or characteristic data of the old and new items were identified to see if the parameter estimates in the hedonic model could be utilized to quality adjust the official price change.

Approximately 54 percent, or 21, of the 39 microwave oven substitute price changes in the study were quality adjusted based on specification differences between substitute items. The most common type of quality adjustment performed for microwave ovens in this study was for changes in capacity—represented in more than 80 percent of the quality adjusted (substitution) prices. Other quality differences such as changes in sensors, type of controls, stage programming, and country of origin specifications were also adjusted for in the hedonic indexes. The table below provides a summary of mean price changes for microwave oven substitutions in both the official and quality adjusted Major Appliances indexes.

Summary of mean price changes for microwave oven substitutions

Substitution price change can be either "pure" (directly compared or quality adjusted) or "imputed" (non-comparable). The empirical results above reveal that not only was all "imputed" price change replaced with "pure" price change in the quality adjusted index but that almost half of the "comparable" substitution price changes in the published index were made "more comparable or pure" in the quality adjusted index. In addition, the mean price changes for each group of substitutions are lower after quality adjustment than in the published indexes. This result implies that some price increases due to quality change may have been reflected in the published indexes and caused them to be higher.

Monthly price indexes were simulated for August 1999 through April 2000 to determine the impact of quality adjusted microwave oven substitution price changes on the Major Appliances CPI. First the published indexes, or without quality adjustment indexes, were recreated by Statistical Analysis System (SAS) computer programs using historical CPI data. The duplication of the published indexes provides a "control" environment from which the quality adjusted Major Appliances indexes can be compared.

U.S. level price indexes, such as those examined in this study, were obtained by summing elementary, or local area, price indexes using aggregation weights derived from the Consumer Expenditure Survey (CE). The results of the index simulation, with and without hedonic quality adjustments, are presented in Attachment 5. In addition, graphical representations of the simulated price indexes for Major Appliances are presented in Attachments 1 and 6.

Attachments 1 and 5 indicate that the U.S. CPI for Major Appliances would have increased 0.6 percent instead of the official index increase of 0.8 percent if hedonic quality adjustment methods had been applied to microwave oven (substitution) price changes from August 1999 through April 2000. Given the small representation that microwave oven substitution price changes had in the Major Appliances CPI during the study period—an average of just four price quotes per month—the impact at the U.S. level was relatively large. In addition, the empirical results suggest that the official index for Major Appliances rose more rapidly than if hedonic quality adjustments had been applied to microwave oven substitution price changes during the study period.

Attachment 6 illustrates the differences between the published and quality adjusted indexes on a monthly basis for the nine-month period in the study—August 1999 through April 2000. Comparisons of the quality adjusted and published Major Appliances price indexes reveal the following: In months when the index increased, the quality adjusted index rose faster than the published index in one month (August), both indexes rose at the same rate for two months (September and April), and the quality adjusted index rose slower than the published index in the remaining two months (November and March). In addition, in the remaining four months when the index fell, there were two months (December and February) in which both indexes fell at the same rate and two months (October and January) in which the quality adjusted index fell faster than the published index.

Two factors may account for the empirical results reported on in this study:

First, the small number of microwave oven substitution price quotes that were quality adjusted may have been to few to make a (representative) impact on the Major Appliances indexes in 1999-2000. Microwave oven substitution price quotes accounted for less than two percent—on average, four of 215 prices—of the monthly CPI sample for Major Appliances from August 1999 to April 2000 and only half of these microwave substitutions had their price changes adjusted using the hedonic technique.

Second, CPI data collection procedures direct BLS field economists to select substitution or replacement items that are the "same or similar" in quality as the old item they had been pricing. This procedure tends to yield substitution price quotes that have fewer bona fide characteristic changes (between substitute items) than might have occurred if the procedures had instructed field economists to collect (substitute item) data for the most technologically advanced or best selling microwave ovens.

This second factor is important in developing expectations as to the direction and magnitude of quality adjusted indexes when compared to indexes without quality adjustments for consumer appliance goods. If BLS field economists were instructed to substitute to the best selling or most technologically advanced microwave oven products, one might expect that the Major Appliances indexes with and without hedonic quality adjustments would diverge further (than in this study) from each other.

BLS is considering additional ways to more quickly bring a greater number of new good quotes into the CPI rather than just relying on the current Telephone Point of Purchase Survey (TPOPS) rotations. Lane (2000) provides a summary of additional methods for bringing new goods into CPI samples more quickly.¹⁵ In particular, both the directed item rotation and directed item replacement methods of updating price index samples instruct field economists to "select a new set of (sample) items representing a more recent period's purchases" for target groups of goods or services that are constantly changing in quality with successive generations of product introductions.

Microwave Oven Characteristics That May Be Important in Future Models

Future generations of microwave ovens could experience rapid quality change if the growth of "smart" appliances becomes popular with consumers. Recent studies indicate that within ten years, 98 percent of (consumer) appliances will have computer processing capability and be networked and controlled from remote locations—such as the office or cell phone.¹⁶ The movement to embed networking capabilities into consumer appliances for the home has been hailed as a major business opportunity for appliance manufacturers, technology companies and (internet) service providers alike. To date, however, these smart appliance products have been costly and used mainly in demonstrations at trade shows and concept homes.¹⁷

Currently, food scientists have created several smart microwave ovens that recognize the food to be prepared and offer advice on recipes and nutrition. In particular, food scientists maintain that the next generation of microwave ovens will incorporate the following "smart" features: ¹⁸

• Microprocessors
• functioning as the "brains" of high-tech ovens. Developed in collaboration with electronics manufacturer's, the microprocessor will control the oven's electronics, such as turning the unit on for the correct length of time and at the proper power setting. It will also be programmed to interpret and analyze food science information, using the expert knowledge base of many food science professionals. The smart microwave can inform you about ingredients and make recommendations to improve your health.

• Bar codes and scanners
• will be the tools for transferring product information into the microwave’s brain. Researchers are working with bar-code scanner makers to develop two- and three-dimensional bar codes, which will be capable of storing considerably more information than the current one-dimensional codes allow.

• Voice recognition

• Internet access
• will keep microwave ovens informed on the latest cooking instructions and nutritional information from web sites. Other internet oven features will include warning consumers quickly about important news such as product recalls; letting the microwave automatically send your grocery order to the supermarket each week; and keeping track of food ingredients consumed for a report to your physician.

Specific examples of "smart" microwave ovens that have been developed to date include the following:

• Sharp
• Electronics Corp. has developed a convection microwave oven that can download recipes from the company's Web site. The microwave then gives step-by-step instructions on preparing the meal and automatically sets the time, adjusts the power and does the roasting, baking, broiling or grilling. Sharp began selling its product in Japan in October (1999) and hopes to introduce it in the U.S. in early 2001. Sharp's new oven sells for about $1,000 in Japan.¹⁹

• Samsung

• Panasonic's
• Virtual Chef Interactive Microwave oven can record a recipe from a television cooking show or download it from the Internet, and then display it later for consumer viewing on a screen. Or, on that same screen, you might input a list of the items you have in your refrigerator and the microwave could suggest a recipe for using them.²¹

• General Electric's
• microwave oven voice-activated Advantium(TM) oven with Speedcook technology that recognizes and responds to voice commands, understands 250 regional accents, can learn new languages, and can self-diagnose problems and alert the GE Answer Center for parts and service.²²

• General Electric's
• microwave oven that reads Universal Product Codes (UPC) and then automatically sets the proper cooking cycle, detects ingredients which consumers may be allergic to, displays the calorie content of the dish, and can self-diagnose problems and alert the GE Answer Center for parts and service.²³

Many manufacturers and retailers of consumer appliances are optimistic that these "smart" microwave ovens—and other "smart" appliances—will become popular with consumers when they start to become available to U.S. consumers in early 2001. Their optimism is based, in part, on "the swift acceptance of the Internet among consumers and because of the soon-to-be-widespread availability of so called 'broadband' internet access, where home users will have swift connections that eliminate the delays and slow downloads of standard modem connections." ²⁴

Many retailers are expected to aggressively advertise smart appliances when they become available in the U.S. economy sometime in early 2001. For example, Sears intends to create room settings inside many of the 860 full-line, mall-based Sears stores to explain to consumers the value-added properties of these new (smart appliance) products and services. In addition, these new smart appliance products will be demonstrated so consumers can relate them to their own living rooms and kitchens. Recently, a spokesperson for Sears indicated that the company "views the Internet as a huge catalyst that will drive transformation in the home, and we intend to position ourselves to play whatever role is necessary." ²⁵

"Currently, the electronics industry is mulling over two emerging, would-be standards for home networking of smart appliances, both based on software widely used in personal computers and across the Internet. One of the home networking standards—known as Universal Plug and Play (UPnP)—is being developed by a team of appliance hardware and software manufacturers headed by Microsoft. The competing platform—a language from Sun Microsystems called Jini—has the same goals as UPnP, which is to provide a standard for networking home appliances. The first generation of both types of smart appliances should be hitting the market in early 2001."²⁶

In short, internet technology is making a whole new line of smart appliances a possibility for consumers. However, these future smart appliances might hold some of the same compatibility problems that computer users endure today. For example, one might be able to turn on an Jini-equipped home security system with a Jini cell phone—but one might not be able to power up an UPnP speaking microwave oven with it. Industry analysts think that there will probably be bridges to connect the two platforms where customers demand it. For example, one might have to work around a compatibility problem by using an Jini-equipped cell phone to turn on your UPnP speaking microwave oven through a personal computer that speaks both languages.²⁷

Technical issues aside, it seems clear that U.S. retailers will be bringing these new smart appliances to the U.S. marketplace for consumer purchase in the very near future. If smart appliances become popular with consumers when they are available for purchase, BLS should try and include these new products in the Major Appliances CPI soon after their introduction so that any rapid price movements—sometimes associated with the introduction of new products—are not excluded from the CPI. Representation of smart appliances in the Major Appliances CPI should be proportional to consumer expenditures on these new products vis-à-vis consumer expenditures on all major appliances. If consumer expenditures on smart appliances increase through time, so should their representation in the CPI.

Conclusion

The microwave oven hedonic model developed in this study represents a snapshot of how the average consumer values quality for microwave ovens in today’s appliance market. Iterative regressions for microwave ovens are included in this study to give the reader a sense of hedonic model development as more categories of CPI characteristics were considered for model inclusion. The current rate of (quality) change for this consumer appliance product is not as fast as other retail products—for example, cellular and wireless telephones or DVD players in the consumer electronics market.

The parameter estimates in the final hedonic model for microwave ovens exhibit relative robustness to the inclusion and deletion of other variables not included in the final model. The direction and magnitude of the parameter estimates seem reasonable, and the statistics pertaining to fit, significance and collinearity are within expected limits. It should be used in conjunction with commodity analyst judgment to quality adjust CPI quote level substitution data when possible.

Future generations of microwave ovens could experience rapid quality change if the growth of "smart" appliances becomes popular with consumers of this appliance. BLS should include these new smart products in the Major Appliances CPI soon after their introduction so that any rapid price movements—sometimes associated with the introduction of new products—are not excluded from the CPI. Representation of smart appliances in the Major Appliances CPI should be proportional to consumer expenditures on these new products vis-à-vis consumer expenditures on all major appliances.

Notes

1. Acknowledgments: the author wishes to thank Nicole Shepler, Walter Lane, Rick Devins and John Greenlees for helpful suggestions.
2. Silver (1998) notes that there are " three different approaches to the use of hedonic regressions for measuring quality-adjusted price changes. The first complements the existing matched models approach generally used by statistical offices, by helping to identify key quality characteristics and, when matches are not available, providing adjustment factors to allow ‘like’ to be compared with ‘like’. The second is the direct method, found in the academic literature, which uses the coefficients on the dummy variables for time in an hedonic regression as estimates of quality-adjusted price changes. The third method requires quite extensive data for the compilation of ‘exact’ hedonic price indices as defined from economic theory."—Page 1 from reference below.
3. Source, "Raytheon Historical Backgrounder," under the Microwave Cooking heading on page 3; <></>
4. Source, "Technology and Economic Growth in the Information Age," National Center for Policy Analysis, under the New Technologies Spread Faster heading on page 6; http://www.ncpa.org/bg/bg147/bg147.html (accessed 3/21/2000).
5. Source, "Toward A More Accurate Measure Of The Cost Of Living," The Final Report to the Senate Finance Committee of the Advisory Commission to Study the Consumer Price Index," under the New Products Bias heading on page 22; http://www.ssa.gov/history/reports/boskinrpt.html (accessed 4/7/2000).
6. Source, Cox and Alm (1997), "Time Well Spent: The Declining Real Cost of Living in America," Federal Reserve Bank of Dallas—1997 Annual Report; see Exhibit 8 on page 22; http://www.dallasfed.org/htm/pubs/pdfs/anreport/arpt97.pdf (accessed 12/8/1999).
7. Source: "Microwave Ovens" in the Consumer Digest’s 2000 Annual Buying Guide, November/December 1999, pages 106-107.
8. For additional information about the hedonics project on quality change in the U.S. CPI see Fixler, Fortuna, Greenlees and Lane (1999), "The Use of Hedonic Regressions to Handle Quality Change: The Experience in the U.S. CPI," Presented at the Fifth Meeting of the International Working Group on Price Indices; August 1999; Pages 1-20. http://www.statice.is/ottawa/bls.rtf (accessed November 1999).
9. The collection of hedonic price data by CPI field economists coincided with the collection of official CPI price data and TPOPS price data. Reports from the field indicated that some respondents simply did not have the time to assist CPI field economists with hedonic data collection.
10. Source: "Microwave Ovens Test: Faster, smarter, bigger," in Consumer Reports December 1999 issue, page 28.
11. Source: Shepler (2000) in "Developing a Hedonic Regression Model For Refrigerators in the U.S. CPI," under the Data and Regression Model https://www.bls.gov/cpirfr.htm (accessed 8/16/2000)
12. Source: "Measuring Prices in a Dynamic Economy: Re-Examining the CPI," The Conference Board, under the Quality Bias heading on page 19; Special Report 1260-99-SR, September 17, 1999.
13. Source: "Microwave Ovens Test: Faster, smarter, bigger," in Consumer Reports December 1999 issue, page 30.
14. Source: "Microwave Ovens Test: Faster, smarter, bigger," in Consumer Reports December 1999 issue, page 30.
15. See Walter Lane, "Addressing the New Goods Problem in the Consumer Price Index," Presented at the Issues in Measuring Price Change and Consumption Conference, Bureau of Labor Statistics, Washington, D.C., June 5-8, 2000, pages 1-26.
16. Source: "GE Unveils Concept 'Smart' Appliances at NAHB," Yahoo Finance—PR Newswire; January 14, 2000, pp. 1-2; http://biz.yahoo.com/prnews/000114/ky_ge_new__1.html (accessed January 19, 2000).
17. Source: "Smart Appliances Hit The Net," TechWeb News / Technology News; January 18, 2000, pp. 1-3; http://www.techweb.com/wire/story/TWB20000118S0032 (accessed May 5, 2000).
18. Source: "New Wave Microwave," Science and Technology News Network^Tm; The Center for Science and the Media, and ScienCentral, Inc.; June 23, 1999, pages 1-4; http://stn2.net/pagesl1/newwave/moreinfo.html (accessed 5/22/2000).
19. Source: "Nukin' meals on the Net," USA Today—Tech Reviews; January 17, 2000, pages 1-3; http://www.usatoday.com/life/cyber/tech/review/crg815.htm (accessed 5/11/2000).
20. Source: "Nukin' meals on the Net," USA Today—Tech Reviews; January 17, 2000, pages 1-3; http://www.usatoday.com/life/cyber/tech/review/crg815.htm (accessed 5/11/2000).
21. Source: " 'Smart' devices could raise machines' IQ," The Daily Camera; January 8, 2000, pages 1-2; http://www.bouldernews.com/business/08delec.html (accessed 5/30/2000).
22. Source: "GE Unveils Concept 'Smart' Appliances at NAHB," Yahoo Finance—PR Newswire; January 14, 2000, pp. 1-2; http://biz.yahoo.com/prnews/000114/ky_ge_new__1.html (accessed January 19, 2000). Also, "GE, Microsoft in pact for talking friges, smart ovens," Yahoo Finance; January 13, 2000, pages 1-2; http://biz.yahoo.com/rf/000113/beg.html (accessed January 19, 2000).
23. Source: "GE Unveils Concept 'Smart' Appliances at NAHB," Yahoo Finance—PR Newswire; January 14, 2000, pp. 1-2; http://biz.yahoo.com/prnews/000114/ky_ge_new__1.html(accessed January 19, 2000). Also, "GE, Microsoft in pact for talking friges, smart ovens," Yahoo Finance; January 13, 2000, pages 1-2; http://biz.yahoo.com/rf/000113/beg.html (accessed January 19, 2000).
24. Source: "Appliances to Be Linked to Internet," Washington Post; January 18, 2000, pages 1-5; http://www.washingtonpost.com/wp-dyn/business/A57379-2000Jan17.html (accessed May 5, 2000).
25. Source: "Appliances to Be Linked to Internet," Washington Post; January 18, 2000, pages 1-5; http://www.washingtonpost.com/wp-dyn/business/A57379-2000Jan17.html (accessed May 5, 2000).
26. Source: "Can Appliances Get Smart?" Washington Post; May 26, 2000, pages 1-3; http://washingtonpost.com/wp-dyn/articles/A8164-2000May25.html (accessed May 26, 2000).
27. Paraphrased from: "Can Appliances Get Smart?" Washington Post; May 26, 2000, pages 1-3; http://washingtonpost.com/wp-dyn/articles/A8164-2000May25.html (accessed May 26, 2000).

References

Consumer Digest’s 1999 Annual Buying Guide

Consumer Digest’s 2000 Annual Buying Guide

Consumer Reports

Fixler, Fortuna, Greenlees and Lane (1999), "The Use of Hedonic Regressions to Handle Quality Change: The Experience in the U.S. CPI," Presented at the Fifth Meeting of the International Working Group on Price Indices; August 1999; Pages 1-20. http://www.statice.is/ottawa/bls.rtf (accessed November 1999).

Silver, M.S. (1998), "An evaluation of the use of hedonic regressions for basic components of consumer price indices," Third Meeting of the International Working Group on Price Indices, Statistics Netherlands, Voorburg: Netherlands (1998) 1-12. http://www.statcan.ca/secure/english/ottawagroup/pdf/23SIL3.pdf (accessed November 1999).

Lane, Walter (2000), "Addressing the New Goods Problem in the Consumer Price Index," Presented at the Issues in Measuring Price Change and Consumption Conference, Bureau of Labor Statistics, Washington, D.C., June 5-8, 2000, pages 1-26.

Attachment 1

Attachment 2

Number and Distribution of CPI Price Changes for Major Appliances (HK01) and Microwave Ovens (HK014 - 01A)

Attachment 3

Iterative Regression 1

Analysis of Variance

Parameter Estimates

Iterative Regression 2

Analysis of Variance

Parameter Estimates

Iterative Regression 3

Analysis of Variance

Parameter Estimates

Iterative Regression 4

Analysis of Variance

Parameter Estimates

Iterative Regression 5

Analysis of Variance

Parameter Estimates

Iterative Regression 6

Analysis of Variance

Parameter Estimates

Iterative Regression 7

Analysis of Variance

Parameter Estimates

Attachment 4

The graph above shows the percent of large and midsize countertop microwave ovens bought new between 1995 and 1999 that were ever repaired or had a serious problem that was not repaired. The data apply to all models within a brand and do not account for manufacturing changes that could affect the reliability of current models. Source: Consumer Reports, December 1999, page 31.

The graph above shows numeric values of the implicit prices, or parameter estimates, estimated for microwave oven brands included in the hedonic model for this product. Brands with negative parameter estimates represent, on average, microwave ovens of inferior quality while brands with positive parameter estimates represent, on average, microwave ovens of superior quality.

Attachment 5

1999-2000 U.S. Level Price Relatives, Indexes and

Index Percent Changes for CPI Item SEHK01, Major Appliances

Attachment 6

Attachment 7

 BUREAU OF LABOR STATISTICS             U.S. DEPARTMENT OF LABOR
 CONSUMER PRICE INDEX - ELI CHECKLIST                                                     
 collection             outlet                       quote            arranging
 period: __ __ __ __    number: __ __ __ __ __ __ __ code: __ __ __     code:  __ __ __ __
 _________________________________________________________________________________________
 ELI No./                                                                     cluster
 title        HK014 MICROWAVE OVENS                                            code   01A
 item availability:    1-AVAILABLE    2-ELI NOT SOLD     3-INIT INCOMPLETE
 purpose of checklist: 1-INIT   2-INIT COMPL  3-SPEC CORR   4-SUB   5-REINIT   6-CHECK REV
 _________________________________________________________________________________________
           CURRENT PERIOD                    |      SALES TAX
                                             |
           price _ _ _ _ _ _ . _ _ _         |      included:        YES    NO
                                             |
           type of price:  REG   SALE        |
                                             |
                                             |                                            
             |
 YEAR-ROUND  |  in-season:  JAN  FEB  MAR  APR  MAY  JUN  JUL  AUG  SEP  OCT  NOV  DEC
 ____________|____________________________________________________________________________
 respondent:                                   location:
 _________________________________________________________________________________________
 field message:
 _________________________________________________________________________________________
 TYPE                                        CAVITY CUBIC FEET
    A1  Table/countertop without             (Full-size models)
          optional mounting                     D1  1.8 cubic feet
    A2  Built-in above the stove                D2  1.7 cubic feet
          or over-the-range                     D3  1.6 cubic feet
    A3  Mounted under cabinet                   D4  1.5 cubic feet
    A4  Table/countertop with optional          D5  1.4 cubic feet
          mounting, mounting included           D6  1.3 cubic feet
          in base price                         D7  1.2 cubic feet
    A5  Table/countertop with optional          D8  1.1 cubic feet
          mounting, mounting has extra       (Midsize models)
          charge                                D9  1.0 cubic foot
                                               D10  0.9 cubic feet
 BRAND                                         D11  0.8 cubic feet
    B1  Amana                                (Compact models)
    B2  Emerson                                D12  0.7 cubic feet
    B3  General Electric (GE)                  D13  0.6 cubic feet
    B4  Goldstar                               D14  0.5 cubic feet
    B5  Kenmore                                D99  Other cubic feet,
    B6  Panasonic
    B7  Samsung                                     ______________________________
    B8  Sharp/Sharp Carousel
    B9  Whirlpool                             MAXIMUM WATT COOKING POWER
   B99  Other brand,                            E1  1100 watts
                                                E2  1000 watts
        ______________________________          E3  950 watts
                                                E4  900 watts
 ** MANUFACTURER'S MODEL NUMBER                 E5  850 watts
                                                E6  800 watts
   C99  ______________________________          E7  750 watts
                                                E8  700 watts
                                                E9  650 watts
                                               E10  600 watts
                                               E99  Other wattage,
 
                                                    _____________________________
 
 
                  
 
 
 ZZ99
 _________________________________________________________________________________________
 BLS 3400B 
                                         HK014 page 1 of 2
                                                                         Revised June 1999
 
    

HK014 - MICROWAVE OVENS - CONTINUED

Last Modified Date: October 16, 2001