4 ms·

The statistics Nate Silver uses are extremely problematic. I would not trust them at all. Any index that uses Fisher price indexes introduces huge amounts of

The statistics Nate Silver uses are extremely problematic. I would not trust them at all. Any index that uses Fisher price indexes introduces huge amounts of subjectivity. For instance, the decision to use an "overlap method" rather than a "direct comparison method" is subjective ( http://en.wikipedia.org/wiki/Fisher_index http://en.wikipedia.org/wiki/Fisher_index ). Choosing one method rather than the other can dramatically alter the resulting index. When you add up a large number of subjective decisions, the resulting numbers simply replicate the assumptions you started with. One government economist writes psuedo-anonymously: "Much of our time is spent 'forecasting,' which basically means making a common-sense appraisal of what some indicator or variable will do in the coming years, and creating a statistical model that confirms it. The second step adds nothing of value to the prediction – the math is just there for show, a means of impressing the innumerate by camouflaging shot-in-the-dark guesses in rigorous clothing." ( http://www.2blowhards.com/archives/2010/03/public_service.html http://www.2blowhards.com/archives/2010/03/public_service.ht... )

cynicalkane 16y ago
The claim that such huge increases in manufacturing can be explained by variations in indexing--well, that's a pretty nutty claim to make. For example, the CPI is actually thought to overstate inflation. And not by very much. The PCE/PI--an index the Fed uses that leaves no room for subjective selection of a goods basket--rises by about a third of a percent less than the CPI. This is a lot in the very long run, but over, say, two decades, that compounds to a little over 6%. Tiny differences in inflation measures--and inflation has been, by any measure, pretty tiny for quite a while--cannot possibly be big enough to invalidate Silver's numbers. The changes are just too big.
- bokonist 16y ago
  Unfortunately I don't have the statistics on hand, but one time I was looking at the average price of cars compared to the price index that had adjustments applied. During a given seven year period, the average price of cars had risen by about 25%. I looked at one specific model (a Corolla) and the price rose by about 15%. But the price index, with adjustments applied, had actually fallen by about 10%. The adjustments applied can make an enormous difference. For example, the CPI is actually thought to overstate inflation. That's not a meaningful statement. Whether CPI overstates or understates inflation depends on the question you are asking, depends on the goods involved, etc. Tiny differences in inflation measures--and inflation has been, by any measure, pretty tiny for quite a while--cannot possibly be big enough to invalidate Silver's numbers. The changes are just too big. Not true, a difference of one percentage point a year compounds, and thus makes a huge difference in long term measurements.
  - mmt 16y ago
    I looked at one specific model (a Corolla) There's a patently false assumption here: that a "specific" model is comparable across a 7-year span. What happened to the price if you normalized on, say, curb weight? With expensive items like airbags, air conditioning, and high-end audio, a car is more like a "basket" than a single item.
rauljara 16y ago
Nate Silver's post has absolutely nothing to do with forecasting. Why include a quote about forecasting? The BLS (bureau of labor statistics) numbers do indeed have some problems with the way they are calculated. Any way you choose to calculate these sorts of things is going to have some problems with it. I don't believe the problems are nearly as big as you seem to think, but that's beside the point. The point is, there are simply no better numbers than the ones the BLS provides. Nate Silver's evidence may have some flaws in it, but there simply is no better evidence out there. There is certainly no evidence out there to suggest that the author of the blog post is correct in saying "we don't make anything anymore".
- bokonist 16y ago
  Nate Silver's post has absolutely nothing to do with forecasting. Why include a quote about forecasting? The quote is a snapshot of the thinking inside government statistics offices. Nate Silver's evidence may have some flaws in it, but there simply is no better evidence out there. Sure there is. Report the actual quantities of how many goods are being manufactured. Look at statistics for how many cars, refrigerators, tractors, memory chips, steel, etc is being produced over time. Then make your own subjective judgement about whether its better to produce a ton of steel or a dozen computer chips.

cynicalkane 16y ago

The claim that such huge increases in manufacturing can be explained by variations in indexing--well, that's a pretty nutty claim to make. For example, the CPI is actually thought to overstate inflation. And not by very much. The PCE/PI--an index the Fed uses that leaves no room for subjective selection of a goods basket--rises by about a third of a percent less than the CPI. This is a lot in the very long run, but over, say, two decades, that compounds to a little over 6%. Tiny differences in inflation measures--and inflation has been, by any measure, pretty tiny for quite a while--cannot possibly be big enough to invalidate Silver's numbers. The changes are just too big.

bokonist 16y ago
Unfortunately I don't have the statistics on hand, but one time I was looking at the average price of cars compared to the price index that had adjustments applied. During a given seven year period, the average price of cars had risen by about 25%. I looked at one specific model (a Corolla) and the price rose by about 15%. But the price index, with adjustments applied, had actually fallen by about 10%. The adjustments applied can make an enormous difference. For example, the CPI is actually thought to overstate inflation. That's not a meaningful statement. Whether CPI overstates or understates inflation depends on the question you are asking, depends on the goods involved, etc. Tiny differences in inflation measures--and inflation has been, by any measure, pretty tiny for quite a while--cannot possibly be big enough to invalidate Silver's numbers. The changes are just too big. Not true, a difference of one percentage point a year compounds, and thus makes a huge difference in long term measurements.
- mmt 16y ago
  I looked at one specific model (a Corolla) There's a patently false assumption here: that a "specific" model is comparable across a 7-year span. What happened to the price if you normalized on, say, curb weight? With expensive items like airbags, air conditioning, and high-end audio, a car is more like a "basket" than a single item.

rauljara 16y ago

Nate Silver's post has absolutely nothing to do with forecasting. Why include a quote about forecasting? The BLS (bureau of labor statistics) numbers do indeed have some problems with the way they are calculated. Any way you choose to calculate these sorts of things is going to have some problems with it. I don't believe the problems are nearly as big as you seem to think, but that's beside the point. The point is, there are simply no better numbers than the ones the BLS provides. Nate Silver's evidence may have some flaws in it, but there simply is no better evidence out there. There is certainly no evidence out there to suggest that the author of the blog post is correct in saying "we don't make anything anymore".

bokonist 16y ago
Nate Silver's post has absolutely nothing to do with forecasting. Why include a quote about forecasting? The quote is a snapshot of the thinking inside government statistics offices. Nate Silver's evidence may have some flaws in it, but there simply is no better evidence out there. Sure there is. Report the actual quantities of how many goods are being manufactured. Look at statistics for how many cars, refrigerators, tractors, memory chips, steel, etc is being produced over time. Then make your own subjective judgement about whether its better to produce a ton of steel or a dozen computer chips.