As long as we are bashing Fox News for being delibatrely misleading, I figure we should bash the White House for doing the same thing, this time court. Macon Phillips, the White House director of "New Media."

So, here's the story:
Recently, Edmunds released a little study they did about the effects of CARS, more commonly known as Cash for Clunkers.

Text below:


Seems reasonable enough. Actually, to me it doesn't sound reasonable, but it is at the very least plausible. The key-takeaways are that Edmunds thinks that most of these sales are sales that would've happened anyways, and so we're basically paying out the $@! for the few additional cars we DID manage to get sold. Basically, this is a subsidy for car companies and a bit for the customers as well.

It doesn't seem particularly venomous, though. Possibly critical, nothing to terribly out of step.


Let's see what the White House has to say...
http://www.whitehouse.gov/blog/2009/10/29/...n-cash-clunkers


Ouch. Accusations of pretty much outright lying, accusing Edmunds of being so out of touch that they are on Mars, calling them stupid poopy-faces...

Not to mention I felt pretty stupid, too. Apparently these results were SO bad that they didn't even stand up to BASIC scrutiny. I read the document and it passed the laugh test (IE, I didn't burst out laughing when I read it), so obviously I must be an idiot. Right? It sure didn't help at all that this guy kept talking about "payback" and 80% of it being in 2009, a term I'm not very familiar with at all. I also felt stupid that Macon kept on ranting on GDP growth and how Edmunds was attacking GDP growth, but I couldn't remember seeing GDP AT ALL in the article.

Am I stupid? I felt like


So I did what any rational person would do when something didn't quite make sense to them: I looked at the sources and thought about the issues. Apparently rational people are hard to find these days, since the common response to something that doesn't make sense is to douse it in fire, burn it, and toss it over a cliff onto a bed of spiky rocks while shouting "HOPE AND CHANGE" or "TEA PARTY RAGE" but that's a story for another day.

The more I look into it, though, the more that Macon Phillips (that's the White House response) looks like a demagogue who is clouding the issues and exaggerating problems with the Edmunds report. Perhaps this is just my interpretation, but it seems he is trying to paint the Edmunds report as ludicrous, completely unfounded, the result of deliberate lies, and should be instantly dismissed out of hand. Ironically, this is probably true, but the issues at hand are severely exaggerated to the point of absurdity, especially the more extreme rhetorical claims that Edmunds thinks car sales takes place on Mars and that the claims don't even stand up to basic scrutiny.

As a full disclosure, I have no idea of Edmunds's previous works on CARS, so I have no idea if their previous work was ridiculous and that this article warranted this kind of extreme response.


The first "problem" is 80% !@#$%^&* with a side of a !@#$ sandwich. In fact, the first problem is a bit more like two problems, so let's break it into 1a and 1b.

1a:


Broken down, what this means is that the Edmunds study assumed that if people had cars that didn't qualify for Cash for Clunkers, they wouldn't buy a new car. Since there were apparently a whole bunch of cars that you could not buy under the CARS program (for instance, you can't buy a Hummer or luxury vehicles), it would follow that sales of these cars were not affect by it. Then you look at sales of THOSE cars, estimate how much the overall market would be based off of that, and then look at the difference between what SHOULD have happened and what DID happen. From that, you can estimate the effects of the overall program.

This is all based on the assumption that Clunkers only affected the market for cars that could actually be bought under the Clunkers program. The more the Clunkers program affects this car market, the more muddled the analysis becomes, making Clunkers look like a worse and worse program. This assumption IS wrong, but it's not clear HOW wrong it is, and the degree IS important. The two markets are linked to some extent and the links aren't always perfectly clear.

Macon makes fun of this wrong-assumption by saying Edmunds is assuming the sales take place on Mars, to emphasize the two market concept. Unfortunatley for Macon, the Council of Economic Advisers report he loves to quote so much in this little report makes no mention of sales increases in the broader market...it talks pretty much only about cars that actually qualify for Clunkers. Apparently the effect he is stressing so much that severely undermines the credibility of the Edmunds study isn't important to enough to warrant inclusion in the CEA. Which does not bode well for his overall argument.

Furthermore, his proposed mechanisms of WHY this relationship would work are a bit suspicious and have no supporting evidence (within the context of this blog post).
His first proposition is price decreases: if people are trading in their clunkers and buying certain kinds of new cars, it should lower the relative prices of OTHER kinds of cars, thus raising sales numbers through an increase in the quantity demanded. Thus, once again, contaminating the analysis.
The relevant problem here is how much the prices changed as a result of CARS and to what extent the car classes compete with each other. "Competition," you see, is a tricky concept, and some classes of goods are more direct competitors with each other: for instance, Xbox competes with Playstation and Legos in the idea that they are all trying to get the attention of kids so mommy buys them at Christmas time. If the prices of Legos fall 10% and the prices of Playstation rise 10%, though, do you think Xbox will have more or less sales? Or if Legos rose 10% and Playstation fell 10%?

The second one is what I call Razzle-Dazzle. According to "reports," people come into dealerships expecting to the rebate from Clunkers to buy a new car, figure out that they are not qualifed, but are SO overcome with all the awesome new American cars that they are compelled to put down thousands for a new one. Car dealerships are apparently uber-dazzling, and enough to motivate people to completely forget about the recession, job losses, and their impaired credit and buy an awesome new vehicle. Needless to say, I am extremely skeptical of the widespreadeness of this phenomenon. I don't doubt that it happens. I also don't doubt that some school children sing songs about how awesome Barrack HUSSEIN Obama is. I do doubt that is very relevant, because, for the most part, car dealerships don't really get me in a shopping mood, and I have a hard time believing that people are making impulse buys at dealerships. Sure, there are "reports." There are also reports that malpractice increases defensive medicine, but I don't think Macon is getting into a huff about that if the political opinions of Obama are any indication.

I don't mean to make wild excuses for Edmunds. These ARE concerning issues with their analysis. But the problems are dependent on how widespread the effects Macon mentions, so the assumption that it is entirely unreasonable to do a study with these techniques is ridiculous. Does it need refinement? Hell yeah, but what measure doesn't? It's ESPECIALLY ridiculous when, at first glance at least, that Edmunds reports HUGE changes in the number of cars of sold. That's to their credit, and shows that their "SAAR" calculation is not terribly unreasonable. That's only a first glance though, so I could be very wrong at that.

Note that this isn't "basic scrutiny" to anyone who isn't thinking about markets once every minute.


The REAL problem, though, is with 1b, or "payback."

The term confused the hell out of me, but that's because I am more familiar with the term "inter-temporal subsitution." In laymen's terms, at least some of these sales were sales that were going to occur anyways, but were going to occur in December or January or February and were moved forward to July and August to take advantage of the tax credit. Makes sense, right? If you have a choice to buy a radio on Friday or Saturday, but the price of the radio is going to be $10 cheaper on Friday, you are more likely to buy Friday.
This is called "pulling forward." The flipside of this is apparently called "payback." When December ACTUALLY hits, your sales numbers are going to be lower since the purchases that were SUPPOSED to be then were in June instead to take advantage of the tax credit.
It's important to note that pretty much ALL sales under the CARS program were pulled forward. These cars are all going to have to be replaced at some point. The relevant question is over what time frame. If you are pulling forward sales from next year, you have a serious problem when next year rolls around. If you are pulling forward sales over the period of 10 years, you have an effective stimulus for the economy and a sustainable policy since you can EASILY pay back 700,000 cars over 10 years.

The Edmunds study assumes that most of the payback occurs in the very near term, which creates a very small increase in sales. This estimate seems to be a pretty clear outlier, as Macon points out. That does not mean it is wrong, though, and the idea of "well other people disagree with you therefore you are wrong" being a "not standing up to basic scrutiny" is laughable and actually a bit offensive. You should attack people based on their data and their interpretations, not because their conclusions challenge the established norms, and to see a White House official engaging in such thinking is pretty damning, even though it's pretty much expected. It's also especially ridiclous, since on Page 15 of the CEA report(remember, that's the one he told us to look up), we find this little gem:


Apparently the CEA doesn't agree with this idea that people who have different estimates should just be ignored and laughed at. Probably a good thing.

Now, once again, I am not making an EXCUSE for Edmunds. The fact that their estimate is an outlier SHOULD mean that people should pay LESS attention to it than they do to other reports unless the entire world gets into a huge buzz about their new awesomeness and innovative measurements. What I reject is the extreme notion that we should entirely dismiss this out of hand, and say that it doesn't even pass "basic scrutiny" and generally debase and disrespect the company that produced the study. "He's different so let's point and laugh at him!" is not change I can believe in.


On to problem 2:


I have no !@#$@#$ idea where this is coming from. I can't find any talk about GDP in the webpage that the White House linked to. The closest I see is this:

I suppose it's possible that this isn't the full press release or that I am missing something, but I think I've read this page about eleven times and I still can't see it. I'd have preferred for the WH man to point out exactly where this statement about GDP was made. But leaving that aside, this argument is 95% !@#$%^&*.

Here's the basic jist:
1. Car companies have inventories
2. They sold lots of cars under CARS
3. They are building more cars to replace their inventory
4. More inventory building means more jobs
5. CARS is good

The problem is #2. According to the Edmunds analysis, #2 is incorrect. Though sales increased this past summer, they are going to fall down through the floor in the future. There was no major shift in sales, and therefore will be no shift in GDP through inventory building. So they're building inventory NOW? Great, all that means is that they'll lay off their workers NEXT YEAR when they find they have too much inventory on hand. No change in GDP. Woot woot!!!!!!!!

Note that I don't actually believe in that, but it follows directly from the Edmunds analysis. So you can't really attack them on this point, because they have a perfectly acceptable counter-argument to it. That makes this "implausible assumption" that was never even MENTIONED not implausible at all. You have to attack them on their analysis, and this doesn't actually do anything to attack their analysis.

EXCEPT FOR ONE PART: The assumption that the car companies building inventories means that the demand MUST be sustainable. This is a !@#$@#$ retarded argument, especially from a lifelong Democrat. Business can and do make missteps all the time, and inventory management can be very tough. Hell, right now the auto industry is running at MASSIVE under-capacity, because of all the recent investments they made for demand that is just not there. So that's why I call it !@#$%^&*.


What he does get right is the following:
"CEA’s analysis is transparent and comprehensive, laying out all of its assumptions for the public to understand. Edmunds.com, on the other hand, is promoting a bombastic press release without any public access to their underlying analysis."
The CEA is pretty open about how they figured stuff out. The Edmunds report is just a press release, which is more less the equvielant of saying "yeah, I did a report, but you can't see it. This is how I did it, and these are my results, but no numbers or specifics for you." That IS slightly worrying. On the other hand, open doesn't necessarily mean correct, so the implication that the CEA analysis is automatically better because of transparency issues strikes me as !@#$%^&*.






And then there's the stuff Macon doesn't even mention. For instance, increased confidence can become self-sustaining and lead us out of a recession, making CARS look better than it otherwise does, and this effect is just as unknown as the connection between the CARS market and the normal auto market. Speaking of market links, the CEA analysis is very flawed in the sense that it looks ONLY at car sales, yet buying a car is probably going to entail cutting back on something else, even WITH the rebate, which makes the CARS program look a LOT worse. Note that this is the exact same problem Macon chews out Edmunds for, and the only leg he really has to stand on.

The GOAL of this piece is laudable: He is trying to get people to not look too seriously at Edmunds and not instantly re-evaluate their opinions on the Cash for Clunkers for program. The MEANS are misinformation, lies of omission, and exaggeration, coupled with some venomous rhetoric. And, unfortunately, it's not the kind of !@#$%^&* that is easily aired out and exposed. It's much more subtle and nuanced, and that's what makes it especially disturbing: it passes basic scrutiny.



Please do not take this as a partisan shot at the Obama administration. I have no idea to what extent Macon's views reflect Obama's, nor do I know how prevalent this is at the current White House or the previous. This is just my opinion about one man's blog post on the White House blog.




Also, do not take me as an automatic defender of all industry studies. The AHIP study that the private insurers put out was crap. This is an individual case that warrants judgement on its own merits.

There seems to be a lot of words to say what everyone already knows: Politics is about misleading people.

This point, however, stands out as almost certainly erroneous:



Considering there has been a massive economic downturn, I suspect seasonal, historic trends are irrelevant. People do not buy cars as often, so saying "10 000 cars were sold per year, on average, in the last 5 years, therefore 10 000 of the 15 000 cars sold this year would have happened anyways" is clearly a faulty method in this case (numbers pulled out of my $@! as an example". Now, I am not an economist like you are, so I might have misread that point, so feel free to correct me.

Well, here is what they did, as far as I can tell:

You have 4 cars, A, B, C, D. C and D qualify for CARS, A and B do not.

Since A and B do not qualify, and shouldn't be impacted much CARS, you can look at historical relationships to TOTAL sales. What they are doing isn't exactly saying "10,000 cars were sold per year on average" but "when people buy 1,000 of A and 1,500 of B, they will buy 10,000 cars overall, according to our relationship that has an adjusted r^2 of .9."

They are using sales of A and B to extrapolate total sales. The UNSTATED question is whether these historical relationships hold up very well in a recession, which is something that's not stated and since Edmunds isn't releasing the data we can't know for sure. The STATED question is whether or not A and B are really unaffected by the CARS program. Their analysis more or less presumes that sales of A and B were gonna be what they were, but the WH blog is saying that A and B are actually significantly affected, and so the analysis isn't so useful.