October 21, 2005
Here and there around the web
Catching my eye here and there around the web: latest housing indicators may not be as bullish as they seem, what to make of all the indictments, price-gouging is all-American, and more insights into fuel use per dollar of GDP.
Housing downturn? Calculated Risk remains bearish despite the strong September value for housing starts and other indicators. He notes:
In the previous housing busts, 'Starts' stayed strong right into the slow down and the excess inventory led to many builders going bankrupt. Usually 'Starts' is a trailing indicator for the housing market.
Indictments. Where would I be without Tradesports to tell me what the news means? Tradesports sez: chances of indictments are 20% for DeLay, 60% for Rove, and 80% for Libby. But probability of confirmation of Miers plunged as low as 20% this morning on stories such as these from Captain Ed. It's since recovered, though still below 50%.
Price gouging. I along with many others have made the point that raising the price of an item in times of emergency is a necessary response to lower supplies and increased demand. To this Christopher Westley adds the following full-throated defense of price-gouging:
The gas station owners, not public authorities, are the ones who risk their capital in order to satisfy customers. They are the ones who hire labor, set contracts with suppliers, and organize resources so as to provide goods to customers via voluntary exchange. They should be able to charge whatever prices they want.
That's why Jason McBride should have been able to charge $5.00 or more a gallon for gas if he wanted. Or he could have given it away for free. Or he could have stacked it up in one-gallon cans, placed a table cloth over it, and had a picnic. After all, it was his property.
Fuel efficiency. Responding to my observation that oil use per dollar of real GDP has steadily been declining, Stuart Staniford notes that miles driven per dollar of real GDP are remarkably constant, leaving one to wonder what magnitude of gas price increase or economic dislocation would be necessary to bring about an actual decrease in miles driven:
Posted by James Hamilton at October 21, 2005 09:05 AMdigg this | reddit
Listed below are links to weblogs that reference Here and there around the web:
» http://www.deadparrots.net/archives/economics/0510.html from The Dead Parrot Society
Stuart Staniford has a post discussing an interesting observation: namely that this vehicle-miles driven data shows steady growth over time. Specifically, when compared to GDP, Stuart derives the following pictures: and relative to 1970 GDP: This data ... [Read More]
Tracked on October 22, 2005 03:42 PM
» Miles Driven, Per Capita from The Dead Parrot Society
I recently agreed with Stuart Staniford and James Hamilton that nationwide miles driven didn't decline appreciably in the face of gasoline price increases that we had experienced in the past. The number of miles driven has continued an upward tack that... [Read More]
Tracked on October 25, 2005 04:04 PM
Are you saying that increasing my driving would be considered a patriotic act, increasing the GDP and thus the prestige of the great U S of A?
Think I'm going to need another vehicle ...
Posted by: MarcV at October 21, 2005 10:44 AM
"eaving one to wonder what magnitude of gas price increase or economic dislocation would be necessary to bring about an actual decrease in miles driven:"
The greatest obstacle to changing Americans' driving habits is not economic - it's regulatory, namely, the outright banning of traditional neighborhood development (dense urban style housing) in just about every housing market in the country. And yes, there is an unsatisfied demand for such housing - one can easily tell by the disproportionately high prices people are willing to pay to live in what few examples of traditional neighborhoods exist (i.e. developed before the advent of single-use zoning after WWII).
If your only housing option is exurban sprawl, you are going to have to drive a lot whether gas is cheap or expensive. Most trips aren't discretionary, and carpooling is not possible in that environment -- the ONLY modification which can be made is to get a more fuel-efficient vehicle, which is hard to do when you first have to trade in your Explorer while everybody else is trying to do the same.
Posted by: M1EK at October 21, 2005 11:08 AM
It is not surprising that the vehicle-miles have a strong correlation with GDP, but it isn't self-evident that the relation should bo practically 1 to 1. Why driving increasis 3% for 3% growth of GDP. Why not 2% or 4%. We might even suspect that the relationship is in fact more tight in reality - calculating GDP has changed since the mid '90s and may overstate the growth.
We should expect the factors behind those miles to be many and complex (cars owned, roads built, employment, urban development, and so on) - but may be this is not the case.
MarcV might have right. The GDP is "made of" moving people and goods, besides making those goods. Spatial activity means economic activity. To work, shop, use services mean moving around in a modern society.
We have here one quite clear component - the population growth of about 1% a year. We can safely assume that this increases proportionately as much GDP as driving. Second clear component is driving itself - the whole system with car making and maintenance, road maintenance, fuel and so on - forms a part of GDP.
But this doesn't explain it all. It must be noted that vehicle-miles are somewhat odd way to look at the transporting activity. Usually we are interested in what transporting is really about - moving people and things around. This means ton-miles and person-miles. I guess that the vehicles here are all kinds - trucks and cars, may be motorcycles, too.
The close correlation could be just a coincidence - structural changes canceling each other out. But probably not. We should check for instance the expanding suburbs. If this is correlated directly to GDP (1 to 1), then this could be an explanation. People put fairly fixed share of their income in housing and get in time more sq. feets for it, and this translates to steadily expanding suburbs. This expansion is close connected to GDP and rather stable for many reasons and doesn't allow great fluctuations in driving. Add population growth and driving-generated GDP component - do we have the explanation?
Posted by: TI at October 21, 2005 11:36 AM
Strong starts sounds like a bullwhip is coming (like boston/mass in the 80s).. I hope the builders have enough coinage to play the beer-game when demand slows. When the marginal cost of building is so much lower that the sales price you have to wonder if builders are going to be looking at too much inventory.
Posted by: Anonymous at October 21, 2005 02:25 PM
According to the BEA, Employed Americans in 1970 stood at 86,946 million vs 139,175 million in 2000. Thus, the 30 year trend is a 160% increase, roughly in line with your growth in GDP and growth in miles driven.
I hate to sound like a broken record (a predecessor of the Compact Disc for you younger readers), but the growth in work force is closely related to the growth in GDP.
The correlation of milage to workforce simply implies that the predominant factor in driving is employment rather than population. Retired people and students do drive, and children need rides, but the difference correlation of miles seems more related to those employed.
Have a happy weekend!
Posted by: Bill Ellis at October 21, 2005 03:29 PM
1980-2003, GDP and miles both grew at average 3.1%. Total population grew at 1.07% and employed population grew at 1.33%. I plotted the data from your link, and the line doesn't match at all. See
Posted by: Stuart Staniford at October 21, 2005 06:17 PM
Is there good reason to believe that the mileage driven statistics can or should be used in this manner?
This chart is from the FHWA data and not the NHTS data, correct? (I speculate that this is the case since the NHTS -- the more comprehensive estimate, I think -- is only collected every 5 years) The FHWA data is, I think, compiled by the states with differing degrees of accuracy and with differing objectives and on different timetables. The FHWA does try to ensure year to year consistency, but that only brings up other potential problems since it presumes the existence of the long-term trend that you are, I think, trying to test for.
I also wonder whether the 2-lane highway mileage data is as accurately collected as the interstate mileage, so the substitution over time from one to the other strictly because of access and/or attractiveness may result in an aberrant overall trend. I'm no expert in these data sets, so this could be way off. Feel free to point that out.
Regardless, it's an interesting angle, I'm just not sure what to make of it.
Given the relative long-term stability and low price of gasoline and given the many marginal cost *reductions* to interstate travel over the years (cars are more reliable, cars are quieter, speed limits are higher, there are more roads, and travel times are likely shorter), I suppose that I shouldn't be entirely surprised at this result.
But the attempt to draw inferences from it has made my head spin. Lastly since this has turned into a Faulknerian-stream-of-consciousness comment, I would also feel better knowing whether there is an incentive to misreport problem with the FHWA data (how much do fed highway funding decisions get affected by state usage statistics?).
Sorry for the length and lack of organization. This is interesting and I appreciate the new angle.
Posted by: Victor at October 21, 2005 07:27 PM
The roadmap data are from Table 3.5 of
and they site this table
which says that "The 50 states and the District of Columbia report travel by highway category, number of motor vehicles registered, and total fuel consumed". I am relying only on this aggregate information. I do not know how reliable the data are (no error estimates are included). I'm just going on the general presumption that official data are accurate enough to be useful, and certainly it's hard to see how all the states could have collaborated to make the miles data match real GDP.
Posted by: Stuart Staniford at October 21, 2005 09:36 PM
"The 50 states and the District of Columbia report travel by highway category, number of motor vehicles registered, and total fuel consumed". Aha. The graph was not at all about vehicle-miles driven. We already know that fuel consumption has grown at the same pace with GDP. Same with vehicles owned. So no conundrum here. Case closed.
Posted by: TI at October 22, 2005 09:02 AM
TI. It is vehicle miles driven. Go look at the reference yourself.
Posted by: Stuart Staniford at October 22, 2005 12:16 PM
Also fuel consumption has *not* grown at the same pace as GDP. The ratio of vehicle fuel consumption to real GDP has dropped by about 1/3 since the 70s. But that's almost all come from vehicle fuel efficiency, not changes in driving patterns.
Posted by: Stuart Staniford at October 22, 2005 12:20 PM
Actually ton-miles per unit of GDP have declined (.59 ton-miles/$ in 1970 to .38 ton-miles/$ in 2002).
We are making more small high value shipments and shifting to a more services based economy with less goods to move around
Which results in growth for Truck miles traveled outpacing the overall increase in Vehicle miles travelled http://ops.fhwa.dot.gov/freight/freight_analysis/freight_story/congest.htm
Posted by: Don at October 22, 2005 01:41 PM
I'm sceptical of the so-called shift to the service economy. Consumption patterns have not changed much at all and 80% of consumption goes on housing, transportation, and food/beverages (according to the BLS CPI surveys). Pretty concrete matter-and-energy based stuff. See:
The way we *produce* those things may have changed, but what we like to consume has not changed much at all.
I wonder how much of the shift in ton-miles has to do with the shift in retailing from smaller local stores to big-box edge-of-town stores. That must have reduced the amount of "last-mile" delivery ton-miles quite a bit (it will have invisibly disappeared into the auto-traffic).
Posted by: Stuart Staniford at October 22, 2005 02:40 PM
The ton-miles per GDP dollar has declined about 1/3 from the '70s. This is same as the drop in fuel consumption per GDP dollar. My point is that the vehicle fuel effciency increase is probably overestimated. Carrying less is not same as vehicle efficiency. Freight transport is only a part of the overall miles but it has significance.
I questioned the vehicle miles data because it seems that it is derived, not measured directly.
Im am quite sceptical to the notion that energy efficiency has increased markedly (somewhat, yes, not not so much) in the US. Imports have embedded energy content that add about 10% to the total energy consumption. GDP accounting has changed and so has the structure of the economy. This makes the comparisons over time difficult. There seems to be a trend break in the '70s and this sould be examined more closely, but after that the changes are so small that error factors may distort the picture.
Posted by: TI at October 23, 2005 05:04 AM
Hmmm. Interesting that since 1997-8 or thereabouts - that is to say, in a period of very low oil prices until 2000 and moderate ones up to 2004 - GDP outgrew mileage quite noticeably, which is pretty much the opposite of what you might expect.
Does this reflect: a) the correlation is a purely statistical artefact? b) in the short term, vehicle use is price inelastic in both directions? c) JIT techniques and computerised stock control led to a more efficient use of freight transport? d) with higher GDP, some of the putative demand for driving instead went to the airlines? e) all of the above?
Justify your answer.
Posted by: Alex at October 24, 2005 02:52 AM
..further to my comment, I'd point out that this phenomenon is pretty much what you'd expect if (as we know) the productivity of a given factor is rising over time. Oil intensity of GDP could be described just as well as the marginal productivity of oil.
Posted by: Alex at October 24, 2005 02:56 AM