I forwarded this the other day, trusting that Merrill Lynch had gotten things right. They hadn't really. On checking with the BEA (which is apparently something the Merrill economist didn't do), I discovered that they stopped publishing the real dollar figure because they wanted to discourage people from adding up the components to get the aggregates, because chain-type numbers aren't additive. (Add them up, subtract from the aggregate, and you get the residual.) They still use the figures internally, and publish a chain-type quantity index for the series they discontinued publishing the real dollar amounts of. The BEA still stands by its hedonic technique.
Doug
Doug Henwood wrote:
>[this is for you data geeks out there - from Merrill Lynch - very
>interesting...]
>
>
>Techonomics
>
>The Government Stops Reporting Real Computer Figures
>Reason for Report: Hedonic Pricing Update
>
>Highlights:
>
>* We previously have pointed out that the government's use of
>hedonic pricing, a price deflator that translates nominal into real
>dollars, appears to overstate technology demand.
>
>* The problem is mostly in the reporting of computer hardware. For
>example, applying hedonic pricing for 2000-03 multiplies the nominal
>dollar increase by eight times to come up with real dollars spent.
>That's money that no vendor saw and overstates real GDP and
>productivity growth, in our opinion.
>
>* The Bureau of Economic Analysis has stopped reporting the real
>computer hardware shipment figure used to calculate real GDP though
>it is still used in GDP calculations. It said, "Commerce was
>concerned the rapid price declines for computers made the figures
>misleading."
>
>* If it's misleading, we recommend that the government improve upon
>its hedonic pricing model to eliminate the exaggerated spending. But
>if the BEA continues to use the current model, it should at least
>report the real dollars and not just the nominal figures. Hiding the
>calculation does not help the users of economic data.
>___________________________________
>http://mailman.lbo-talk.org/mailman/listinfo/lbo-talk