More on adjusting for quality when calculating inflation
In an earlier post on inflation (remember, inflation is supposed to capture only price changes, not quality changes), I put forward the notion that the official statistics from the ONS may actually understate inflation in certain products when they incorporate quality adjustments in their calculations.
I've done a little more digging and here are a few notes on quality adjustments in the consumer price index (CPI):
An ONS paper by Adrian Ball and Andrew Allen provides a good overview of quality adjustments made to measured inflation. The authors note that until January 2003 quality adjustments were made using a the 'option cost' method:
'In option costing the retail cost associated with a change in specification is obtained from the cost of purchasing the change separately or as an added option. Fifty per cent of this is added to the price of the original model to give a price comparison that is independent of any changes in quality.This method of capturing and accounting for quality improvements in products in measured inflation seems overly crude, in my opinion. Fortunately, from February 2003, the CPI started using a 'hedonic' approach for certain technology products. The word 'hedonic' relates to the concept of derived pleasure of the individual and the 'hedonic regression method' is generally used for high technology products where upgrades are frequent, as it accounts for changes in measurable characteristics of products such as computer memory and processor speeds, digital camera photo resolutions etc. For technical details on how this quality adjustment is calculated see here (a nice, short note on hedonics) and pages 47-49 of the CPI Technical Manual .
Fifty per cent of the cost is applied for a number of reasons. In part this is because 100 per cent option cost could lead to an over adjustment for quality change, due to the fact that the cost of buying features separately is generally greater than buying them as a package.'
Both documents provide similar examples where the hedonic calculation is applied to a computer whose price was measured in January but has seen a processor upgrade in the following month. In the second example, while the actual price has increased by 32%, after the hedonic calculation the price is shown to fall by 4.4% (that is, after stripping out the change in quality). It's quite the difference.
To conclude, after all this digging, it does indeed look like the ONS may be overdoing it. In my earlier blog post, I gave the example of replacing my laptop five years later. After applying the hedonic adjustment, it would surely show appear that the price of my laptop would have plummeted over five years. My view is 'that it (my laptop) really hasn't fallen that much (in price) at all. Why? Simply because my old laptop is now obsolete. It simply not relevant to compare the new laptop to the old laptop. The philosophical point relates to the question of 'what exactly is a laptop?'. I agree that you do get more for your money in the short-term. However, as time progresses this gain quickly evaporates to zero.' I may be missing a crucial point here (?) but it does seem as if something is amiss and that the official statistics may actually understate inflation rather overstate (as most articles and economics textbooks claim). Indeed, Ball and Allen conclude in their paper that, 'For the HICP, the downward impact on the computer price index was more marked though because the weight associated with PCs in the HICP is small the change still would have had little impact on the overall published index.' I believe this is an important issue that needs to be resolved because the role of technology in our lives is continually expanding to encompass ever more products.
