Wages have averaged flat for the last 20 years "with inflation", but $1000 for medicine buys you a much better product today than the "equivalent" amount of money 20 years ago. Traditional basket of goods models tend to assume quality parity.
In longer time frames food has maintained value for very long time periods, but salt has dramatically dropped in value. If you compare baskets the prices that stay drown out the ones that drop off a cliff. Which makes inflation a tricky subject over very long time periods.
EX: Food and land in ancient Mesopotamia has value in our modern economy. Clay tablets are kind of meaningless, and pottery is really hard to judge.
Basically, I completely agree that it is tricky, and traditional hedonistic adjustments really are pretty poor approximations.
As to clothing, I have bought cotton clothing that barely survives a single wash. Sure it's cheap, but it's like they started designing clothing to look good in your closet if you never wear them. It feels like your choices are 50 loads for 60$ or 2 +/-2 loads for 5$.
PS: The sad thing is even orange juice has gotten worse. Fresh or frozen makes little difference they both suck.
He's noted that life expectancy improved 2x more from 1900 to 1950 as from 1950 to 2000.
The whole matter of hedonic adjustment is another hairball. It omits a few elements.
Maslow's hierarchy of needs. Quality of life is most enhanced by fundamental aspects: food, shelter, clothing, and safety. You cannot build the pyramid from the top.
Quality can be measured various ways. Product lifetime, utility, repair requirements, etc. There's been shifts in multiple directions, some better, some worse.
Fungibility. Products aren't infinitely miscable collections of features. Smartphones offer connectivity but the user pays the piper in surveillance and privacy loss.
Points are that:
1. LE hasn't increased much post 1950.
2. LE at a given age hasn't increased much post 1950.
3. Where LE has increased, it's largely been in minority / underserved populations, presumably as a result of improved basic care or access.
4. QALYs show little improvement post 1950 / 1970.
Healthcare expenditures have increased greatly. They've offered little actual benefit, other than to healthcare companies and providers.
#3 Iceland for example is at 81.8, #22 UK 79.4, #33 United Arab Emirates 78.7, #38 United States 78.2
so does the CPI and every other inflation measure using 'hedonic adjustment' ... new BMW comes out, new price, you look at it and estimate how much is price change and how much is quality change, bigger car with air bags, voice activated GPS, self-steering and whatnot...imperfect but necessary.
The CPI, PCE, and related indices tend to approximate the GDP deflator but are generally inferior as there is no sound way to perform the hedonic adjustment or to choose baskets in the first place. Eric Weinstein has proposed gauge theory as a way to handle substitution effects but has not, to my knowledge, computed an index using it.
The weights are different (reflecting the different weight in the GDP vs. the CPI reference basket, also depending whether you refer to the implicit GDP deflator, the fixed-weight deflator, or the chain-weight deflator.)
Also, to compute real GDP you additionally have to estimate prices for a lot of things that aren't in the CPI, but might be in the PPI, import prices etc.
GDP just takes all the other stats on prices, consumption, investment, government spending, brings them into a consistent system of national accounts. Nothing generally superior or inferior, and all these stats including a hefty dose of judgment and a large margin of error.
http://www.bea.gov/papers/pdf/NABEspeakernotes.pdf
There's always someone will argue it's unsound, just as someone will argue the merits of one programming language over another, but the people who make the judgment calls are pretty well-versed in what they're doing and do the best they can.
if you think the GDP contribution of owner-occupied housing, i.e. imputed rent, corresponds to a physical quantity proportional to entropy, more power to you!