Likewise, when phones with the Android Ice Cream Sandwich operating system appeared, they commanded a premium over phones with Android Gingerbread, the predecessor operating system once its successor-Android Jelly Bean-appeared, Ice Cream Sandwich phones sold at a discount. Thus, one might expect a positive coefficient on 3G early in its life cycle and a negative coefficient once 4G phones are available.

For example, a particular mobile telecommunications generation (2G, 3G, 4G) will represent relatively high quality when first introduced and relatively low quality once the succeeding generation enters the market. We next consider the impact of allowing the effect of these characteristics on price to vary over time using the adjacent-quarter regressions in Eq. The regression results indicate this set of variables explains 83.9% of the variation in price. As shown on line 4, the price index constructed by chaining the coefficients on the quarterly dummies falls 19.6% per year, on average, from 2010Q1 to 2018Q1. To recap, the first hedonic regression we report includes quarterly time dummies, variables for engineering capability, dummy variables for manufacturer and operating system, and a dummy for the quarter a model is introduced in the market. Whatever the correct explanation, we find that adding a dummy variable for the first quarter a model appears in the market has a noteworthy impact on our results as discussed below. 25 In addition, Apple typically introduces new models in September-the third month of the third quarter- implying that the average price observed in the IDC data for that quarter is not an average of 3 full months of price history as for other quarters, and that difference may distort estimates of the price change in the following quarter. 24 Or, very early buyers may demand a discount in exchange for bearing the burden of testing the new features of the phone. 23 Alternately, the market may be in disequilibrium at first in the sense that the “law of one price” does not hold and the new item is sold at a lower quality-adjusted price than incumbent models (put another way, the price of the old model does not fall enough to equilibrate to the price-performance ratio of the new model). In particular, prices paid when models are first introduced may be affected by a number of factors that confound quality adjustment: When a model first appears, the novelty of the item may lead select early-adopter consumers to pay a premium for the item that does not persist over time. Our rationale for including an introduction-quarter dummy is to account for price variation over the model life cycle that does not merit inclusion in measured quality-adjusted price trends. We also include a dummy for models in their quarter of introduction. 22 Specifically, we include dummies for phone manufacturer (there are 43 in our dataset), for operating system (there are 58 in our dataset, for example, “Android Froyo 2.2”), for telecommunications generation (2G, 3G, 3.5G, 4G), and for input type (QWERTY keyboard vs. (17.1), regressing (the natural log of) price on four continuous measures of engineering capability (storage capacity, screen size, camera resolution, and processor speed, all in natural log form) and an array of dummy variables. Sichel, in Measuring Economic Growth and Productivity, 2020 17.4.2 Hedonicįor our hedonic indexes, we first consider the simple, time-invariant specification in Eq. The combination of improved data availability and new (or revisited) methods for completing first- and second-stage hedonic analyses promises to keep hedonic valuation in the forefront of nonmarket valuation research on the environment.Īna Aizcorbe. Newer approaches in second-stage analysis require somewhat fewer data than the traditional multimarket approach and show substantial promise for broadening the use of second-stage hedonic estimation. In this case, a second-stage analysis is needed that adds information on households to estimate underlying demands or utility parameters. Furthermore, the first-stage hedonic model is sometimes insufficient to recover WTP for an environmental amenity, or a researcher is specifically interested in how demographic characteristics relate to WTP. Specification of the hedonic model, measurement error, omitted variables, and endogenous regressors are some of the major considerations, and each topic is the subject of active research. The hedonic model, while a powerful and frequently used tool in nonmarket valuation, has many considerations that must be addressed when attempting to identify the marginal WTP for an environmental amenity through observations on housing prices. Taylor, in Encyclopedia of Energy, Natural Resource, and Environmental Economics, 2013 Conclusions

Hedonic price