Do Price Deflators for High-Tech Goods Overstate Quality Change?

This paper studies chained price indexes for goods in high-tech sectors, how they handle quality change and whether they will likely suffer from chain drift issues. We argue that chained indexes do not handle quality change properly; though the underlying bilateral indexes hold quality constant just fine, the chained index does not, a problem we call comingling. When we explored multilateral methods as alternatives to chained indexes, we found that the GEKS index that is often used to deal with the chain drift problem does not fix the comingling problem. Thus, the comingling issue is not just the chain drift problem in another guise; it can also exist in indexes that do not suffer from chain drift. Moreover, the practical implication of this finding is that we cannot use the GEKS index as benchmark against which to judge the numerical importance of these issues. Using the Weighted Time Product Dummy (WTPD) index as a benchmark, we show that, the changes in prices and expenditure shares over the product cycle that are typically exhibited by these high-tech goods can generate numerically important chain drift problems. Moreover, we show that the gap is negative: that is, the chained index will show faster price declines than the WTPD index. We illustrate these points using scanner data for six categories of consumer electronic goods.