Guest Post by Willis Eschenbach
Inspired (as I often am) by either the insights or the foolishness of a guest post at Judith Curry’s always-provocative blog, I decided to take a look at the relationship between fuel price and miles driven. My inspiration came from my amusement at the guest author’s use of the following graph to establish a relationship between fuel cost and how much people use their cars. I think a relationship exists, but the graph used by the author doesn’t show it. Figure 1 shows that graph:
Figure 1. Per capita fuel use, compared to the fuel price, for the OECD countries. SOURCE
Now, it certainly looks like there’s a clear relationship there, but that’s an illusion. My objection to the graph was, the countries divide into two groups. On the bottom right you have the European OECD countries, plus Japan. Plus one fish.
On the top left, you have the US, Australia, Canada, and New Zealand. What’s not to like?
Well, ignoring fuel price for the moment, who would you think would drive more miles—a citizen of the US, or a citizen of Japan? An Aussie, or a Belgian? A Canadian, or an Italian? So all the guest author has shown in that graph is that the folks in large countries, with miles and miles between cities, drive more than Europeans and Japanese.
But of course, I couldn’t leave it there, so I linked to the following lovely graph of automobile use in the US that I ran across during my research. It shows, year by year since 1956, how many miles Americans have driven, and what the gas price was during that year.
Figure 2. Miles driven compared to the fuel price. Click to embiggen. SOURCE
Now that shows some very interesting patterns. The main oddity I noticed is that there is what might be termed a price shock effect—in the year of a big jump in prices, for example 1974, the mileage driven drops compared to the previous year. But then look what happens from 1974 to 1978 … the price stays stable, but the number of miles driven each year goes up steadily, without reversal.
But of course, I couldn’t leave it there. I digitized the data, to see what kind of relationships I could understand and reveal through further analysis. And as usual, I was surprised by what I found.
First, taking the data as it is given, there is no statistically significant relationship between the two variables, pump price and miles driven. The R2 is only 0.03. (“R2” is a measure of the relationship between two datasets, where an R2 of 1.0 indicates a perfectly linear relationship between the two, and an R2 of 0.00 indicates no relationship. So an R2 of 0.03 is … well … pathetic. So as far as a direct relationship between prices and miles driven, not happening.
Once I saw that, I wondered, well, what if I include a temporal trend in the linear regression? The way that I usually do that is simply to include the date as a variable. And to my surprise, the R2 went from 0.03 up to 0.98 … Figure 3 shows an emulation (multiple linear regression result) of the number of miles that Americans drive, versus the value estimated based on year and pump price.
Figure 3. The emulation is a multiple linear regression, using the year and the pump price as independent variables, and the actual average miles driven by Americans as the dependent variable. R2 = 0.98
Dang, sez I … that’s pretty impressive.
But of course, I couldn’t leave it there. A fixed annual increment, a simple trend like I used, is just a way to understand the data. It’s not an explanation involving some plausible mechanism. And more to the point, I also didn’t like those two years up at the top right of Figure 3, which are 2009 and 2010. In those years, Americans drove about a thousand miles less than expected. So I though about why that might be, and even a bear of little brain would go “global financial meltdown, duh”. And that made sense overall as well, because how far I drive doesn’t just depend on the pump price. It also depends in part on how much money I have in my jeans. When I’m flush I drive more, and when times get hard, I drive less regardless of the price of gas.
So I thought that instead of using the year, I’d try using the per-capita GDP as the second independent variable. Figure 4 shows those results.
Figure 4. The emulation is a multiple linear regression, using the real per capita GDP and the pump price as independent variables, and the actual average miles driven by Americans as the dependent variable. R^2 = 0.98 GDP SOURCE
Yowzah! Now that’s what I call shaving with Occam’s razor. It turns out that pump price and per capita GDP do an excellent job of estimating the number of miles driven, with very little error.
So, what does the magic equation that gives us the excellent results shown in Figure 4 say about the relationship between miles driven on the one hand, and gas price and per capita GDP on the other?
Well, it says that for every twenty-five-cent increase in the pump price of gas, Americans drive about a hundred miles less. Gas price goes up, miles driven go down. Makes sense.
And it says that for every $430 increase in per capita GDP, Americans drive about a hundred miles more. Wealth goes up, miles driven goes up. Also makes sense.
Now, the “carbon taxes” I’ve seen discussed are on the order of $20-$30 per tonne of CO2. And by coincidence, $28 per tonne of CO2 emitted is equal to twenty-five cents per gallon of gasoline. So if a $28/tonne carbon tax is imposed on gasoline, how much less might Americans drive?
Well … a hundred miles less … wow, such a stupendous gain, be still, my beating heart …
And how much actual change in our driving habits is a hundred miles less per year?
Well … since Americans drive about 10,000 miles per year, it’s a gigantic, massive reduction in miles driven of one percent.
And that, dear friends, is all the bang you get for your twenty-five-cent per gallon carbon based energy tax. A one percent reduction in miles driven. One freaking percent, and they want to impoverish the poor for that? Grrrr ….
So … what does this mean for the debate on carbon-based energy taxes?
First, it means that in the American situation, there is no way that the benefits of energy taxes are worth the cost. Why? Because the effect of a typical CO2-based energy tax on miles driven is minuscule, only a 1% reduction for a $28 per tonne of CO2 energy tax.
Next, a very slight increase in per capita GDP will nullify the energy tax entirely. Also by coincidence, it turns out that if the current per capita GDP goes up by about 1% (~$430), that will increase the mileage driven by 100 miles … so a 1% increase in per capita GDP will completely nullify a $28 per tonne of CO2 energy tax. And the GDP goes up by one percent all the time …
Next, it means that in order to have more than a one-year effect, the tax will have to continually rise.
The problem with a carbon based energy tax can be seen by thinking back to Figure 2, where I noted the “shock effect”, and how after the slight reduction in miles driven as a result of the 1974 big jump in pump price, after that one-year reduction the miles driven went right back to increasing year after year, with no change in the gas price.
So a one-time jump in the price will make little difference, just a one-year reduction in the miles driven. But by the next year or two, assuming that the per-capita GDP continues to rise as it has in the past, the miles driven will be rising again.
Next, it means that a carbon-based gasoline tax is wildly regressive. To see why, let me start with a slight digression, by bringing in a concept from accounting, that of “fixed”, “variable”, and “semi-variable” costs.
Fixed costs are those costs you can’t do anything about. The amounts are fixed, you can’t reduce them, you just have to pay them.. Maybe rent. Taxes.
Variable costs are costs that are entirely optional. Think maybe eating at restaurants. You don’t have to spend a penny on that if you don’t want to.
Semi-variable costs are costs that you can change, but you can’t eliminate entirely. These would be things like food costs. You can run them up or down, but you can’t eliminate them.
Now, think about the corresponding concepts as applied to the subject at hand—fixed, variable, and semi-variable miles driven.
Fixed miles are things like a commute to work. Short of changing your job or your residence, you can’t change that. You just rack up those miles every year.
Variable miles are things on the order of visiting Grandma in Arizona. You love to do it, but you don’t have to go.
Semi-variable miles are things like going to the post office to get your mail. You can cut the trips down, but not to zero.
What this graph shows me is that any energy tax on gasoline will hit the hardest on the poorest, the people who mostly use their car to get to work. The problem is not just that more of the wages of the poor go to energy, although that is also a problem.
But in addition to the higher percentage of their wages going to energy, the majority of their miles are fixed miles, so they can’t cut back on them. They have to drive them, so they have to pay the tax.
For the wealthy, on the other hand, lots of their miles driven are variable or semi-variable, so they can just scale down a bit. The energy tax means nothing to them. But for the poor, it can be a budget-buster.
This is one of the many reason why energy taxes are so regressive—because for the poor, fixed costs for everything squeeze them all the time, not just fixed fuel costs but also the other bills they have to pay every month. So when energy prices go up, Al Gore and James Hansen just cut back on visiting the grandchildren they love to talk about, no problem for them.
But the single mom whose gas budget barely covers getting to work, she can’t cut back on her gas use, it’s already cut to the bone. So when she pays the energy tax, she is forced to cut back on something for either the kids or herself.
And all of that for a pathetic 1% reduction in miles driven. That’s criminal.
Now please, folks, don’t insult my intelligence by claiming that it’s OK to harm the poor because of that well-worn fantasy, the fabulous claim that wealth redistribution will make it all OK. It won’t. Anyone who believes it will make it all OK has not spent enough time around government programs.
To start with, even the best-intentioned programs only reach a percentage of those most affected. Next, the poorer that people are, the less likely they are to hear about such programs. Think people living in apartments versus people living in their cars. Next, the paperwork required is all too often complex, confusing, and intrusive. Next, many of the poorest people are mistrustful of government. Also, immigrants are often equally fearful of government, and many don’t speak the language. Next, the people who end up getting the most benefits are often not those who suffered the most losses. Next, administering such a program requires a large expensive workforce of bureaucrats and paper pushers to make it function. And of course, they’re all Union, can’t be fired, plus we’ll be stuck paying these pluted bloatocrats their megabucks in retirement money ’til they shuffle off to a warmer place … and I’m not thinking Florida. Next, as with any government program, waste will consume more than you imagine. Think IRS conferences in Las Vegas and thousand dollar hammers. Next, parasitic rent-seekers like lawyers and consultants will be circling the honey-pot and making off with some of that good honey. And finally, there’s never been a government program that people didn’t scam, game, and cheat, so somewhere between a little and a lot of money will simply be stolen.
So no, wealth distribution will only make things worse, or on the best day with a following wind it might “break even” by taking from one bunch of the poor and giving to another bunch … and meanwhile the people at the bottom of the economic pile are hit the hardest. And whether you are a conservative or a liberal, that should appall you.
And finally … we’re going to create all that pain and create a giant bureaucracy and waste piles of money for a crappy 1% reduction in miles driven, a temporary reduction that will be wiped out by the next 1% increase in per capita GDP?
Really? That’s the brilliant plan? Screw the poor and the economy for a 1% reduction in miles driven?
Spare me. That’s more than foolish, that’s a crime against the indigent and everyone else in the country. Almost any other conceivable response to the imagined horrors of CO2 would be preferable. Taxes on energy are destructive and damaging to individuals, to businesses, to the environment, to the economy, and more than anything to the poor, and to turn it from mindless idiocy to criminal tragedy, there is nothing to show for it at the end of the day but a temporary 1% reduction in miles driven—from an energy tax, there’s no lasting gain, only lasting pain.
w.
DATA: The spreadsheet with the data and graphs is here.
[UPDATE] I just wondered, how much will the $28 per tonne of CO2 gasoline tax cost per year? Average fuel economy of the US fleet, cars and trucks, is about twenty mpg. Average person drives ten thousand miles, at twenty mpg that’s five hundred gallons. The tax at twenty-five cents per gallon on five hundred gallons is $125 per year.
In response to that tax, we can expect people to cut fuel use by 1%, or 5 gallons per year. Gas is around four bucks a gallon, so that’s $20 worth.
So the plan is to charge the average driver $125 per year in gas tax, and in response to that he’ll use $20 less gas, reducing his bill at the pump from $2,000 per year to $1,980 per year and cutting his CO2 emissions by a whacking great 1% … who thinks these plans up, and how can we catch them and stop them?
[UPDATE 2] I also got to wondering, just how much CO2 would a $28 per tonne of CO2 applied to gasoline consumption actually save? There’s 8.9 kg (19.6 pounds) of CO2 in a gallon of gasoline. Crazy but true, it’s the extra weight of the oxygen. So we’d be saving one whole percent of that, or .089 kg per gallon. Multiply that by the number of gallons of gasoline burned in the US, about 134E+9 gallons, and we end up with 0.01 gigatonnes (billion metric tonnes, E+9 tonnes) of CO2 saved.
And compared to a hundredth of a gigatonne, how large are the global CO2 emissions? Well, it’s about 9 gigatonnes of carbon C emitted per year, so as CO2 the mass is (16 + 16 + 12) / 12 of that to allow for the extra weight of the oxygen, or 33 gigatonnes of CO2 per year.
And the $28 carbon based energy tax would reduce that by 0.01 gigatonnes of CO2, which is a reduction of three hundredths of one percent (0.03%) … folks, have we truly gone so mad that such a trivial gain, three hundredth of one percent reduction in CO2 emissions, so small as to be absolutely unmeasurable, is used to justify this crazy tax?
“John says:
July 10, 2013 at 6:53 am
Of course they would like to change with me, because I’ve made decisions to give me a choice.”
And what does one do when one does not have that choice?
Willis, I disagree… but not with the analysis per se. It is true that higher energy costs reduces energy use very little. And when it does it is not always good – think of all houses with too much insulation and to little ventilation that is infested with mould and mildew. But anything you tax becomes dearer and on the margin less of it will be used. Raise income taxes on the poor and more people will be unemployed, raise an energy tax and less energy will be reduced. Even a minimal government will have to tax something and I prefer less environmental damage to more unemployment.
I also would add that preferably a tax is taken as late as possible in the economic cycle. Energy taxes are both a production tax (when it is used for e.g. production or commercial transport) and a consumption tax (when it is used for e.g. residential heating). In my eyes it makes it neutral and a good revenue source that internalizes environmental damage (i.e. makes it cost) and strikes at least some part in the later stages of the economic cycle.
(If you’re going to argue that governments can’t be trusted to lower income/payroll taxes if they are allowed to raise energy taxes, well, yeah, could be…)
“Joe Born says:
July 10, 2013 at 6:59 am
Now, you may argue–and I may agree with you–that some prices affect the poor more than others; it’s at least plausible that increasing the cost of, say, milk or sugar will have a greater impact on the poor than raising the Tesla’s cost.”
I do believe you have never had to chose, heat or eat (Or anything else actually)!
Willis: Of course, the real impetus is simply to generate more tax revenue, regardless of the excuse.
Mike M,
You may do whatever You want unless You affect others. It is not a freedom You talk about, it is anarchy. Your freedom end where starts my freedom and vice versa.
I intentionally stated “If we can agree”, those “we” in USA should be citizens of USA. You asked: “WHO gets to make the determination whether or not a given act is wrong?” Answer is simple, the society. And every member of society is rights to do the best to convince others.
Ptrick,
“And what does one do when one does not have that choice?”
There are always a choice. Just in many cases we don’t like the alternatives.
John says –
“I wanted to show other perspective, but it appears that nobody is interested. Very sad.”
Frankly John, it sounds like you live in a bubble, out of touch with reality. Some perspective eh?
John says:
July 10, 2013 at 6:53 am
So you think all the poor choose to be poor by making poor decisions. Try telling that to the millions who are living from paycheck to paycheck because the cost-of-living has been going up faster than their income,or those that have lost jobs, or have been foreclosed on their house.
another hit to the poor is that our food is transported by truck trains and boats so the added taxes for carbon are added to the “food chain” increasing food prices on everything.
I completely disagree with the premise, that the original graph doesn’t show what it claims to show, i.e., gas prices are inversely correlated with per capita fuel use, the implication being that gas pricing impacts the amount of fuel used per capita.
Of course it does. Try raising prices to $100 per gallon and see what happens to per capita fuel use, even in the larger countries.
But the relationship goes both ways, because you’re really looking at the result of a supply/demand situation in each country.
In the U.S., politician’s heads would be on the block if they tried to put taxes at EU levels. Why? Because the impact on their constituents’ lifestyles would be extremely oppressive due to the way that the transportation system has evolved here. In other words, low gas prices permit long driving distances, so people choose to live far from where they work, and will take their car for convenience even if a bus runs alongside them to work each day.
In other words, the U.S. has a gas pricing structure that more closely approximates a true “free-market” price, simply because the politicians don’t dare tax gas much more than they already do. Even the relatively low tax that is levied is primarily used to improve the road systems in most states. EU countries have perverted the free market by determining that their people need to change their behavior, so they’ve taxed gas (and cars) oppressively. They’ve done this for so long, and so consistently, that the people have adjusted without a revolution, but make no mistake, they have adjusted.
When politicians decide to tax something on the pretext of changing their constituents behavior for the better, you can bet on two things: 1) They don’t really give a rat’s butt about your behavior, and 2) They want more money to spend on stuff they really do care about. In the meantime, their tax increases pervert the free-market pricing structure, which usually leads to unintended consequences elsewhere in the economic system.
Public Transit is largely designed to take advantage of economies of scale. Whether Public Transit can work largely depends on how people and their destinations are organized. If people “live” in one small area and “work” in another small area, then transit is reasonably efficient. However, best case 50% of the vehicle miles are essentially “empty” as you must return the vehicles to pick up the next load of passengers.
If however passengers and their destinations are essentially random it is a nightmare to design an efficient transit system. You cannot achieve the economy of scale necessary to fill and route Transit vehicles such that you can pay the driver and vehicle costs. Your costs per passenger mile go through the roof, as do passenger wait times.
It is a nonsense to suggest that public transit can solve this problem, because it is hugely expensive to supply Transit in these circumstances. To solve the random problem you need many more smaller vehicles. However, the labor costs of having a paid driver for each of these vehicles quickly overwhelms the costs of the vehicles and fuel, making it much cheaper and more efficient overall if people drive themselves.
When cities are laid out such that passengers and their destinations are largely random, then private vehicles are significantly less expensive and more efficient, less time is wasted that could be used more productively, and overall CO2 will be reduced as compared to using Transit. Otherwise the transit system essentially becomes a very inefficient, expensive taxi service that provides poor service because it doesn’t drive you door to door.
This is what most people miss when they suggest Transit as the “universal” solution to private vehicles. The model only works when demand is concentrated. When demand is diffuse, private vehicles are much more efficient, because you don’t have empty vehicles driving around waiting to pick up passengers. With private vehicles the vehicles only move when they have passengers.
John,
True. I could.
I live on what passes for a mountain around my neck of the woods, about a thousand feet above the floor of the Tennesee valley, but let’s ignore that. It does rain an average of one out of every three days where I live. It gets pretty darn hot in the summer, and pretty darn cold in the winter, but let’s agree that I ought to just ignore that too, that I ought to butch up and deal with that.
Of course, I have kids to get to school. Kids to pick up from school and transport to after-school care. I have Dr.’s appointments to get them to, dentist appointments, hockey games, karate classes, voice lessons. I have groceries and other goods to buy.
There is no public transportation within any reasonable distance of my work, my home, or most of these destinations where I live.
No thanks.
Watchman is right about the population density issue. If you have a big country but everyone lives in a few small spots the population density of the country will be low, but the average person will live in a high density area. Perhaps something like the median population density weighted by inhabitants averaged over a grid formed from 25sqmi blocks over the country. Thus Siberia would be weighed much lower than Moscow.
Miles per gallon has been increasing so that the use per capita does not quite match the same graph.
Additional note – countries with the low miles driven and high taxes are also all below replacement birth rates. We should never adopt policies that lead to not enough babies. Really, who is going to feed all the 70 year olds in Russia in 15 years?
“John says:
July 10, 2013 at 7:19 am
There are always a choice. Just in many cases we don’t like the alternatives.”
Well, in my experience, in Africa, people walk. They don’t even have the luxury of a bicycle, let alone rods to use it on! So, no roads, no bicycle, no car, no public transport, no beast and cart, what are the alternatives?
The UK tory party solved the problem way back, realising that a one off fuel price increase had only short term effect they introduced the “fuel price escalator”. Fuel tax increased each year by the rate of inflation plus four percent. This is why we now have the world’s highest fuel prices.
Wow, those people in the big cities don’t drive much do they?? Willis, I average much closer to 40,000 miles a year. Many you could expand your analysis a bit more, exclude those that live in the top twenty cities, and see what the average mileage is for the rest of us!
How can there be “8.9 kg (19.6 pounds) of CO2 in a gallon of gasoline” when, according to http://wiki.answers.com/Q/How_much_does_a_gallon_of_gasoline_weigh, gas weighs 6.073 pounds per US Gallon. Just askin’.
John is what I call a person of little thought. He sees the world from his own vantage point and believes that what is right for him applies to all. But I think it is worse than that. He believes that he knows what is best for all of humanity. Our greater prosperity and longevity is because we continue to master the use of energy and it has never been the case that energy use has been metered out by overlords of society for maximum benefit with the exception of wartime rationing. Energy is everything and more energy is better than less energy. If there are problems, there are solutions. I think that Elon Musk ( SpaceX, Tesla, PayPal, etc. ) would not claim to know what is best for all of us. What he says is that he thinks he knows a better way and puts his time, talents, and treasure behind proving that to be true. John is just a clanging gong with nothing to show for his life but words. ( I am striving to not have my words be my only legacy ). Actions speak louder than words John.
Perhaps Elon Musk can get behind clean and safe nuclear technology, safe pebble bed reactors, or thorium reactors to complete the revolution of energy abundance with maximum environmental benefit. More affordable clean energy would benefit all. Battery technology is getting better all of the time. But you can’t legislate innovation and prosperity. I have looked down on the valley in LA and seen the brown smog. It is much better than it use to be and it is getting better all of the time. But socialist forcing cannot compete with capitalism. Just look at what Kennedy did with the private sector. Get the EPA and other government organizations out of the standards business, set the long term goals, lower the cost of doing business, and get the government out of the way of the private sector. Get the Johns out of our lives.
“So you think all the poor choose to be poor by making poor decisions.”
No, they didn’t choose to be poor as companies don’t choose to bankrupt. But people may have made poor decisions.
However true reason why African are poor is their lack of capital (including human capital). However most poor people in Western World are poor because of bad decisions (for example, no attention to subjects in school, too much partying, using drugs, as well spending money where the money shouldn’t be spent.
Richard Howes says:
July 10, 2013 at 7:36 am
How can there be “8.9 kg (19.6 pounds) of CO2 in a gallon of gasoline” when, according to http://wiki.answers.com/Q/How_much_does_a_gallon_of_gasoline_weigh, gas weighs 6.073 pounds per US Gallon. Just askin’.
——————–
Cause the oxygen that the engine burns the fuel with doesn’t come with the fuel.
Molecular weight of CO2 / atomic weight of C = 44/12 = 3.66667
6 * 3.666667 = 22 pounds, ballpark of the CO2 in gasoline. Of course gasoline isn’t pure carbon that converts 100% to CO2.
Richard,
Let’s consider Methane CH4. When you burn it you make 1CO2 and 2H20. The CO2 weighs more than CH4 (almost three times as much). Same basic process, you grab oxygen out of the air and add it to the carbon. The inclusion of that oxygen raises the weight.
“John says:
July 10, 2013 at 7:39 am”
(for example, no attention to subjects in school, too much partying, using drugs, as well spending money where the money shouldn’t be spent.”
You are talking out of a hole in the back of your head! There are more degree holders in Addis Ababa, Ethiopia than in Sydney, Australia!
Yet another example of “John” opening mouth and changing foot!
A carbon tax on fuel will also impact commercial vehicles, like those involved in delivery of essential goods and services. Think food delivery, clothing, household goods and the like. A fuel price increase will naturally be passed on to the consumer in higher prices for those goods, thus increasing the attack on the poor.
In other words, the stated ends (cutting carbon) don’t justify the meanness.
John,
A lack of capital is not the reason Africa is poor. Their capital level a couple hundred years ago was not significantly different than that of the rest of the world. Additionally they do not have significantly fewer natural resources than other parts of the world. Something else is different between Africa and Europe.
Poverty may be largely caused by poor choices. However, taxpayers paying someone at the IRS $100,000 per year to work for a union that argues for IRS agents who then enforce a tax that raises the cost of gasoline disproportionately felt by the poor doesn’t help. If I have an extra $200 in my pocket at the end of the year I will either spend it or save it. If I spend it someone somewhere gets paid for contributing to the economy since I buy a good or service. If I save it someone somewhere gets hired as the bank lends the money to an employer. If it gets confiscated by the government to raise the price of gasoline, what contribution is made? What good or service has the IRS agent on “official time” provided? The single mother working at IHOP (my mom growing up) is better off if I have that extra money and use it to go out to eat than she is if that money goes to the IRS worker.