Just Gas?

Chris Bowers discusses the implications of the chart above, which appears to show the price of gasoline (mapped inversely so that the line goes down when the price of gas goes up) exerting a dominant influence on George W. Bush's popularity. I was super-impressed the first time I saw that chart, but now I'm not so sure. Aside from the fact that gasoline has generally gotten more expensive and Bush generally gotten less popular, are we really seeing a correlation here? There must be any number of quantities that have also generally moved in one direction during the relevant time period.

Consider the quality of the basketball teams fielded by the New Jersey Nets. Like Bush, they were at their best in the season immediately following 9/11 and have been in slow but steady decline since then. But the Nets aren't exercising a causal influence on Bush's popularity. What's more, I think the price of gas is being demarcated in nominal terms here, which is clearly the wrong way to do it.

Comments

One reasonable hypothesis is that Bush's popularity and gasoline prices are both correlated to events in the Middle East.

Posted by MattF | October 7, 2007 11:24 AM

I do think I saw a study once which showed that movement is gas prices was highly inversely correlated with Presedential popularity, over a several decade span, but I may me not be remembering it correctly. If true, it really suggests the difficulty in getting carbon taxes implemented.

Posted by Will Allen | October 7, 2007 11:28 AM

Uh, no, events in China and India have far more to do with the current price of gasoline, along with refining bottlenecks in the United States.

Posted by Will Allen | October 7, 2007 11:31 AM

"Demarcated in nominal terms"? What the fuck does that mean?

Posted by Anonymous | October 7, 2007 11:34 AM

are we really seeing a correlation here?

Yes, you are a very strong one. The causal connection is unclear, though. I suspect conditions in Iraq (and maybe New Orleans as well) are confounding variables, as MattF points out.

Events in China and India may have a larger effect, but that doesn't mean the correlation with Middle Eastern events isn't a strong one.

(Forgive me for being pedantic, I'm a stat teacher.)

Posted by Matt Stevens | October 7, 2007 11:38 AM

Gosh, I'm no fan of the Carter Presidency, but I've long suspected that his cratering popularity was mostly due to events outside his control, and if gas prices hadn't skyrocketed, the hostage fiasco was something he could have weathered. Ronald Reagan and Bill Clinton greatly benefited by having gas prices tumble during their Presidencies.

Posted by Will Allen | October 7, 2007 11:43 AM

Gosh, I'm no fan of the Carter Presidency, but I've long suspected that his cratering popularity was mostly due to events outside his control, and if gas prices hadn't skyrocketed, the hostage fiasco was something he could have weathered. Ronald Reagan and Bill Clinton greatly benefited by having gas prices tumble during their Presidencies.

Posted by Will Allen | October 7, 2007 11:43 AM

When people get that frown on there face when discussing Bush, gas prices don't seem to be at the front of there complaints. A wasted war we'll do that.

Posted by Pioneer10 | October 7, 2007 11:45 AM

Spurious correlation from co-trending data is a pretty well-known problem in time series analysis. That said, it looks to me that you could de-trend both time series and still see a correlation. In other words, it's not just the downward slope of both lines that's striking, it's the within-the-slope, up and down jumps that seem to fit together so well.

Posted by matt_c | October 7, 2007 11:46 AM

Re Carter

In the case of James Earl Carter, far important in his fall in popularity then gas prices was gas lines.

Posted by SLC | October 7, 2007 12:05 PM

I think nominal, as opposed to real, prices may be the correct measure, if the causal relationship is an emotional one. In my experience, people tend to hold the price of gas out as a special figure, which they don't often relate to generally rising prices.

Posted by Matthew | October 7, 2007 12:17 PM

Statistics 101:
Correlation does not prove causation

Posted by Duncan C. Kinder | October 7, 2007 12:19 PM

it's not just the downward slope of both lines that's striking, it's the within-the-slope, up and down jumps that seem to fit together so well.

Yeah. I doubt that high oil prices are causing Bush's low approval ratings, but I'd guess that both trends have an underlying cause: the ongoing mess in Iraq creates both uncertainty in the futures markets and an unpopular president. So some of the minor fluctuations to which Matt C refers probably do reflect real-world news events, both in Iraq and elsewhere.

Posted by James Gary | October 7, 2007 12:19 PM

Excellent piece by Chris Bowers, and essential reading. The paragraph beginning:

"The days of super popular administrations..."

should be dispositive. Very weak working majorities in apocalyptic conditions are actually likely unsustainable, and process liberalism will not survive. There will be no majorities or coalitions to be found, and Vanguardism/dictatorship, syndicalism, or anarchism are the options. Iraq is our own future.

Jesus ignore this correlation quibbling, and just read the Bowers post.

Posted by bob mcmanus | October 7, 2007 12:22 PM

Statistics 201:

A very strong correlation which exists over a long period of time, and which is buttressed by other empirical data, like Presidential popularity generally being affected by economic conditions, is not to be dismissed outright as being merely the outcome of random chance. There is a decent chance that this is not the same as the rooster crowing prior to sunrise.

Posted by Will Allen | October 7, 2007 12:26 PM

I agree with Matthew (non-Yglesias) - nominal prices probably are the way to go in this case.

Posted by Mario | October 7, 2007 12:30 PM

To Anon 11:34am who wonders what the F nominal variables are, try http://economics.about.com/cs/macrohelp/a/nominal_vs_real.htm

Posted by nbt | October 7, 2007 12:40 PM

Statistics 301:

I can believe that the gasoline market aggregates information in a way that's not too different from the way that presidential approval does... but that still doesn't prove causation. There has to be a model, a mechanism, a something. 'Rising gas prices make people unhappy' is not a model.

Posted by MattF | October 7, 2007 12:54 PM

Someone have mentioned that "wasted war" is more on people's mind than gas prices.

Even so, why is the war wasted? Honestly, the situation in Iraq is bearable, the number of killed soldiers seems to taper down lately, the surge "worked" etc. So why such negativism?

One possibility is that surprisingly many people believed that this is "war for oil", and some did not like that idea from the start, while others actually liked it. However if the war for oil does not bring oil, something is wrong.

People who believed that war is a good thing did it for variety of reasons, and the only thing that somehow panned out was that we are fighting terrorists there so we will not have them here. In lo! no terrorists here. But there was supposed to be also oil, and there is none, breaking the spirit of the "mullahs", who are more cheeky than ever, oasis of freedom that people are fleeing from by millions etc. To a midly clueless Joe Q. Public, this lack of oil from the war for oil is very apparent anytime he pumps gus, and given miserable efficiency of his vehicle and commuting distances, he does it often.

Then both Bush and Cheney are oil people and they started their reign pretty much claiming "what is good for ExxonMobile is good for the country". Recall Cheney energy plan? And sneering at the idea of promoting energy efficiency? To many Americans it sounded convincing: secure incentives and rely on free market, (perhaps add a well-selected strategic war) and enjoy cheep gas forever. Clearly, "it sounded" belongs to the past.

Consequently, as they are patently false on securing the flow of oil, the need for energy conservation, the method of cutting "mullahs" down to size etc., the anti-terrorism achievemnt is not credible.

One thing is that gas prices are painful to the "suburban base" which became something rather than something (two long Brooksian words, but basically they regain the faculty of critical thinking unlike some evangelical die-hards).

Posted by piotr | October 7, 2007 1:00 PM

Consider the quality of the basketball teams fielded by the New Jersey Nets. Like Bush, they were at their best in the season immediately following 9/11 and have been in slow but steady decline since then.
Why does Jason Kidd hate America?

Posted by Gold Star for Robot Boy | October 7, 2007 1:03 PM

I'm curious as to why gas priced tumbled not long after 9/11. Did we open up the strategic reserves? I don't remember that we've ever done that.

Posted by DonBoy | October 7, 2007 1:07 PM

I find it very interesting that big drops in gas prices (that is to say, jumps in the blue line) happen a) right after September 11, b) right around the time of the invasion of Iraq, and c) just after Hurricane Katrina hit -- all events that people thought would drastically increase prices. In the case of Katrina, it appears that thought actually materialized as a pre-storm-and-immediately-after increase in prices which very quickly corrected. Note also that the post-Katrina decrease nicely precedes a clear Bush bump.

Also of note, d) the couple months before the November 2006 elections, and e) a very brief decrease in October 2002.

Posted by Aaron S. Veenstra | October 7, 2007 1:31 PM

Aaron, what you're noticing is that the gas price level can be heavily affected in the short term by releasing some of the Federal emergency reserves.

What people should know is that Democrats could enforce roughly 15% lower gas prices (should they gain serious power in 2008) by legislating the creation of greater refining capacity. This would also remove some of the spikes often seen in the US gas market.

Posted by Meh | October 7, 2007 1:39 PM

"...could enforce roughly 15% lower gas prices (should they gain serious power in 2008) by legislating the creation of greater refining capacity."

I didn't know new refineries could come online so fast. And I also didn't realize that Congress could mandate a fairly risky multi-tens-of-million dollars private industry investment. Good news, if true, we'll just force Exxon to go solar.

Just nationalize the fuckers.

Posted by bob mcmanus | October 7, 2007 1:53 PM

This post is just bait for me. I don't see the comparison to the Nets at all.

The Nets regular season wins since 01-02 are: 52, 49, 47, 42, 49, 41. So two seasons ago was as winning as 5 seasons ago. Bush has been steadily under .500 for quite a while. If you want to go by post season, it is: Finals, Finals, 2nd Rd, 1st Rnd, 2nd Rd, 2nd Rd. Also doesn't track Bush, who's been out of the playoffs since early 2005.

Posted by Al | October 7, 2007 2:22 PM

Matt, I have a hard time you'd pursue such an absurd argument. I could grant an argument that said, for instance, that gas prices are only one of the general facts that effect most people in the U.S. That would be a real argument for the fact that there is only soft correlation. I could also see an argument that gas prices are symptomatic here, not causative - that just as the Iraq war has driven down Bush's popularity, they have driven up gas prices.

But your argument from the NY Knicks is laughable. So what - that there are several trend lines that could possibly parallel Bush's fall in popularity isn't a valid argument that some trend lines correlate to his fall in popularity and some don't, it is an argument that you don't understand how to pick out correlating trend lines. Period.

Actually, I think you do, and that you are presenting this argument sophistically.

Posted by roger | October 7, 2007 2:25 PM

There's a correlation all right...but in the case of the last 7 years, the correlation of gas prices and presidential popularity results not from causation from a shared cause behind both: US gov't policies against the interests and wishes of the majority.

Posted by heatkernel | October 7, 2007 3:12 PM

If I could, I'd run the graph again and put health insurance premiums in for gas.

Posted by eRobin | October 7, 2007 3:18 PM

One speech by the President of the United States pledging the government's total resources to finding alternative energy sources to replace crude oil would send reverberations throughout the world oil markets. And by that I mean a beneficial effect on consumers.

Posted by AGK | October 7, 2007 3:30 PM

Insurance premiums wouldn't be nearly so pretty. They only move once a year, and there ain't big signs up on every street corner trumpeting them. People's sensitivity to gas prices far outweighs it's actual significance because we buy it, all by itself, every week. I'd also just like to agree that adjusting for inflation would be kind of crazy, when looking at political impact rather than economic. I mean, would you expect people to say, oh, well, it isn't just gas, it's general inflation, so that's all right then?

Good to see you've been trained to instinctively quibble about unadjusted price trends, though.

Posted by Boring Commenter | October 8, 2007 8:36 AM

What matt_c said.

Posted by Kyle | October 8, 2007 12:29 PM