Bias and Inefficiency in Quality-Adjusted Hedonic Regression Analysis

Abstract

Numerous quality-adjusted hedonic price-trend studies based on computer prices have provided support to widely held suspicions that officially released price indices are not accurately measuring the price declines occurring in many information technology (IT) products. If verifiable, then general price inflation is being overestimated and, consequently, real GDP is being underestimated. Existing evidence, however, is inconclusive. First, empirical findings for IT-products other than computers are extremely rare and, secondly, estimation bias is inherent in the hedonic regression technique most commonly employed. This paper presents an unbiased method together with an estimated quality-adjusted price trend for laser printers (1993-2004).