When do Matched-model and Hedonic Techniques Yield Similar Measures?

Abstract

Hedonic techniques were developed to control for quality differences across goods and over time in order to construct constant-quality aggregate price measures. When the available data are a panel of high-frequency data on models whose characteristics are constant over time, matched-model price indexes can also be used to obtain constantquality price measures. We show this by demonstrating that, given data of this type, certain matched-model indexes yield price measures that are numerically close to those obtained using hedonic techniques.

Hedonic Price Indexes for Personal Computer Operating Systems and Productivity Suites

“…Among issues prominent in the recent literature on hedonic price indexes are the presence and interpretation of temporal instability in estimated coefficients of hedonic price equations (e.g., Berndt, Griliches, and Rappaport [1995], Berndt andRappaport [2001, 2002], Aizcorbe [2003a, b], Heravi and Silver [2002]; Pakes [2003], and Triplett [2003], chs. 3 and 5); factors affecting differences between price indexes computed using matched-model compared to hedonic regression methods (e.g., Aizcorbe [2003b, c], Aizcorbe, Corrado, and Doms [2003], Heravi and Silver [2002], and Triplett [2003] ch. 5); choice of functional form (e.g., Berndt and 7 A useful recent discussion of econometric estimation issues in hedonic regressions in general, and with a specific application to personal computer hardware, is that by Pakes [2003].…”

“…Rappaport [2002], Diewert [2003], Feenstra [1995], Pakes [2003] and Triplett [2003], ch. 5); and the use of fixed-effects econometric specifications instead of explicit product characteristics measures as regressors in hedonic price equations (e.g., Aizcorbe, Corrado, and Doms [2003]).…”

“…For completeness, we also specify and estimate a fixed-effects ("Fixed Effects") variant of the Quarterly Average model, where each distinct product (e.g., Microsoft Office 2000, Microsoft Windows 98) is assigned a separate dummy variable, and while annual and quarterly time dummy variables are also incorporated as regressors, none of the quality attribute measures is included as a regressor. This Fixed-Effects specification has been advocated by Aizcorbe, Corrado, and Doms [2003], who argue that when prices are observed at relatively short time intervals (e.g., monthly or quarterly rather than annually), and when each product's price is observed frequently, then a fixed-effects specification without quality attribute variables will approximate rather closely the traditional hedonic specification.…”

Life Cycle Price Trends and Product Replacement: Implications for the Measurement of Inflation

“…The studies, such as Boskin et al (1997) and Lebow and Rudd (2003), which provide estimates of the overall bias in the CPI, did not explicitly take account of life cycle bias. In contrast, a great deal of attention has been given to identifying and quantifying quality change bias in the CPI (see, for example, Aizcorbe et al, 2003;Pakes, 2003;Aizcorbe and Pho, 2005;Hill and Melser, 2008;Bils, 2009;Broda and Weinstein, 2010;Erickson and Pakes, 2011;Greenlees and McClelland, 2011;Fox and Melser, 2014). Using data on 29 supermarket product categories included in the Dominick's scanner data set, we find that the life cycle component of price change is indeed significant.…”

“…() and Lebow and Rudd (), which provide estimates of the overall bias in the CPI, did not explicitly take account of life cycle bias. In contrast, a great deal of attention has been given to identifying and quantifying quality change bias in the CPI (see, for example, Aizcorbe et al ., ; Pakes, ; Aizcorbe and Pho, ; Hill and Melser, ; Bils, ; Broda and Weinstein, ; Erickson and Pakes, ; Greenlees and McClelland, ; Fox and Melser, ).…”