In reply to Benjamin Khan: As I said in my first comment on this subject,  
GDP and related concepts are used by governments and private research  
agencies to study such macro variables as inflation, employment.,  
unemployment and exchange rates. For these purposes, economic theory and  
evidence tells us that it is market transactions that matter. If, e.g.,  
unpaid workers work more, ceteris paribus, this will not add to inflationary  
pressure. But if paid workers work more, earning more income which they  
spend, this will add to inflationary pressure.  
  
So if we had it to do again, we would have a measure that was restricted to  
the money value (in current and constant dollars) of transactions that flow  
through markets, removing e.g., imputed rents for owner occupied housing.  
Then we would have, as we do now in many cases, a series of satellite  
accounts than measure other things in which we are interested, such as the  
market value of household work (not easy to estimate as some of the  
exchanges on this topic illustrate), environmental degradation, resource  
depletion, etc  But for most things the departments of finance and of  
industry do (whatever they are called in various countries), it is the money  
value of transactions that flow through markets that is the relevant  
variable.  
  
It is essentialist to ask: What is "the" best measure of national income?  
What one needs to ask is: What is one interested in? and then design a  
measure that best gets at that. GDP tells us the value of what goes through  
markets.  Other measures are much better at judging national "welfare" or  
living standards and many text books, including my own, make that point in  
detail with illustrations such as 'home heating is provided free in hot  
countries and so in not included in the GDP (but does contribute to living  
standards), while it is costly in cold countries and so is included in the  
GDP".  
  
Richard G. Lipsey