Wednesday, August 13, 2008

Official Statistics Questioned

Via Wille Faler, I see this interesting video explaining how government statistics is distorted. I don't agree with all of it (more specifically, his critique of imputations is misguided, and I also find it implausible that real inflation is as high as he claims) but it is still worth watching as he also makes a lot of good points, by for example pointing out how "hedonics" is assumed to be a one way affair, even though real life quality changes aren't.

9 Comments:

Blogger Jim Driscoll said...

OK, I'll bite: why is it OK to impute the rent that homeowners pay themselves and include that in the GDP? And free checking? Or is it some other aspect of imputations that you are objecting to?

But I'll agree: even as someone very, very skeptical of government statistics, the ShadowStat statistics don't look to be any more reliable, and those are what he was relying on for his True CPI numbers.

I'd love to see someone set up a futures market for a real CPI gauge like Case-Shiller did for the housing market - it's not a coincidence that the official gov't housing numbers aren't nearly as severe as Case-Shiller, but Case-Shiller is all that's quoted in the media now.

1:06 AM  
Blogger stefankarlsson said...

Jim: "why is it OK to impute the rent that homeowners pay themselves and include that in the GDP?"

Because it represents the value you gain from not renting it from others-or renting a house you own to someone else. That represents consumption of housing services that home owners in effect bought when they bought the house.

"even as someone very, very skeptical of government statistics, the ShadowStat statistics don't look to be any more reliable, and those are what he was relying on for his True CPI numbers."

Yes, that's why I wrote that these numbers are implausibly high (I do think inflation is underestimated, but not by as much as 8%).

8:20 AM  
Anonymous Anonymous said...

Beräknas inflationen på samma vis i Sverige också nuförtiden?

9:36 AM  
Blogger stefankarlsson said...

I princip ja, men förmodligen dock inte i fullt lika stor utsträckning.

10:05 AM  
OpenID banjokossan said...

Woppidoo! I hope someone does something similar for Swedish numbers sometime. It would be interesting to see what numbers one would arrive at, even if they aren't totally reliable.

8:40 PM  
Blogger Jim Driscoll said...

"Because it represents the value you gain from not renting it from others-or renting a house you own to someone else. That represents consumption of housing services that home owners in effect bought when they bought the house."

You've explained what it is, but not why it should count toward overall economic activity, despite not actually being an actual economic activity - only an implied activity.

OK, let's just spell this out: If I own a house, then the fictitious value of the rent that it may (or may not) be worth will be counted as part of the total economic activity of the country, despite the fact that no such economic activity has taken place.

However, if I similarly own my car free and clear, the amount that it would or would not rent for will NOT count against that same number. This despite the fact that "that represents consumption of transportation services that car owners in effect bought when they bought the car".

I could make other analogies, such as my kitchen functioning as a restaurant, and my suits returning a rental cost every time they're worn, but I trust I made my point.

Why are houses treated differently? Why count something that doesn't actually exist, while not counting other things with similar non-existent characteristics?

Also, don't we count the house purchase once already? Aren't you then counting this shelter activity twice?

I'm of a mind to agree with the video, since this all seems too convenient, but since you don't agree, (and you're trained at this), I'm sure you have a really good reason. I very much want to hear it.

8:44 AM  
Blogger stefankarlsson said...

Jim, actually the purchases of houses are not included in GDP, so it is not really counted differently. What you may have in mind is residential investment spending, but this number is based on the cost of constructing homes. This is basically not different from how non-residential investment include the various costs needed to make candy bars or cars.

You can argue of course that the same could be applied to other consumer durables, such as cars, and in a way that would be correct.
However, because houses are usually far more durable (often lasting for a century or so) than other consumer durables which usually last for only a few years or at most decades, there is a case for treating it differently.

In any case, over the long run, it wouldn't really make that much difference for aggregate GDP. Because if you take away the imputed value of rent, you have to add the full value of house sales. Or alternatively if you impute the rented values for other consumer durables, you have to take away the sales value of these consumer durables.

6:11 PM  
Blogger Jim Driscoll said...

actually the purchases of houses are not included in GDP, so it is not really counted differently.

This was my basic misunderstanding, thanks for pointing it out.

In any case, over the long run, it wouldn't really make that much difference for aggregate GDP. Because if you take away the imputed value of rent, you have to add the full value of house sales.

Thank you! That answers the question for me in full.

However, isn't it true that rents, and thus imputed rents, have disconnected sharply from home sale prices for the last 10 years? If anything, that then means that GDP was *understated* for the last few years, and will be *overstated* for the next few years as prices correct, and rents return to a more historically expected ratio...

The rent/price disconnect isn't small, either, so this isn't a margin of error kind of thing - it's the same objection that one can make to the horrible OER stat in the CPI.

Or am I again misunderstanding the basic elements that go into the GDP?

6:53 PM  
Blogger stefankarlsson said...

Jim, you do have a point about the rent/house price disconnect, which is why I wrote "over the long run". During the housing bubble, house prices rose a lot faster than rents. During the latest year, house prices have fallen while rents have continued to rise. And it seems pretty safe to say that rents will continue to rise relative to house prices during the coming years.

And yes, that means that nominal GDP growth was understated (assuming that the purchase approach is the right one) during the boom, while it will be overstated in coming years. However, since this means that inflation was understated in the past while it will be overstated in the coming years, it will not affect real GDP.

9:21 PM  

Post a Comment

<< Home