Mustang Parts
   Carrying Saleen wheels and Bullitt wheels.

Wednesday, May 27, 2009

No Obvious Political Bias In Chrysler Dealership Closings

Update: Blogger Marla Singer at Zero Hedge did some actual statistcs on the entire dealer open and closing list, and found a small correlation between open dealers and donations to Hillary Clinton's campaign, but not a very significant difference between Democrat and Republican giving overall (0.93 vs 1.05 odds ratio of being closed, D vs R).

The blogger Doug Ross did some digging, and discovered that very few of the Chrysler dealers on the closing list gave political contributions to the Democrats in the last election cycle, and most of the ones that did give gave to Republicans.

In his original post (here), Ross found that out of a collection of 39 dealers he checked, 4 gave half or more of their contributions to Democrats (10%)

As much as I would like to believe it to be true, I don't think it is. There is a major flaw with his analysis: he didn't bother to check the dealers that are staying open to see how their political contributions broke down.

My hypothesis is that Chrysler dealers, being small businessmen, are more likely to donate to Republicans than Democrats, for predictable reasons. Like any small businessmen, car dealers want lower taxes, a lower minimum wage, fewer regulations, etc.

The list of Chrysler dealers that are staying open is here. This is a list of the dealers whose contracts will be assumed by the new Chrysler-Fiat.

What I did was go down the list and copy names into the search box at until I got a clean hit. I did this until I got 25 dealers with donation records for the 2008 election cycle. The results can be considered randomized, since we don't expect a correlation in the alphabetical name of the dealership with their political affiliation (or geographic location).

Result: 23 dealer principals donated to Republicans, 2 donated to Democrats, or about 10%.

If you run these numbers through a binomial distribution, you get a 2-tailed P-value of 1! That means that the hypothesis that the two distributions are identical is correct. Chrysler dealers (and probably all auto dealers) who donated for the 2008 election cycle overwhelmingly supported the Republicans.

Here is my "raw data", if you want to spot check me.



Bumby said...

I saw that, too, and I thought that the argument was a stretch. Thanks for doing the math.

Anonymous said...

You'll need to dig into all the data to confirm any suspicions (in either direction).

Also you will need to look at individual cases to see why a dealer was closed. It's a safe bet to say buddies of the democrats are at less risk of being closed; however, that doesn't mean that they are targeting repulican dealers. But if they are protecting their buddies that is wrong and needs to be highlighted.

Anonymous said...

Why not overlay the shuttered dealers against a list of who didn't buy cars during the 140,000 unit phone beg? The plot may thicken....

Anonymous said...

Thank you for your analysis. I am currently studying for the LSAT and this is a prefect example of logic they test. Good to see the real-life implications.

Anonymous said...

An Engineer Elsewhere says:

"The results can be considered randomized"

That is dubious. I'll reserve judgement pending something a bit more...comprehensive.

That and a public airing of the actual criteria used in the selections.

Or how about a data slice of *profitable and productive* dealerships being closed vs political donations. Nevermind the question why and profitable dealerships would be closed anyway...

Anonymous said...

Why would the government want PROFITABLE dealerships to close? Especially in this economy! It makes no sense at all-they are the ones providing jobs.
Fiat had no interest in cutting dealers, so it was the "task force" that mandated it.

Anonymous said...

The big 3 have been wanting to cut down on their number of dealerships for many years, as have many of the import nameplates. Both Chrysler and GM are taking advantage of their current situation to do that which they've wanted to do for a long time.

There are numerous economic reasons for wanting fewer dealerships. One is that each dealer needs a certain amount of "showroom" stock for each weekend's impulse sales, so more dealers mean more stock that's sitting around waiting to be sold. Yes, the dealers eat some of this cost, but the manufacturers have to subsidize it in various ways.

Another factor is that multiple dealers mean more price competition amongst the dealers, which drives the average actual sales price down. When the average price goes down, the dealer network pressures the manufacturer for more kickbacks.

A third factor is that there is an overhead with maintaining a relationship with each dealer -- shipping, management, etc. -- and if you can cut your network by a 1/4 you can have a big savings.

So, what this means is that while most of the dealers who were cut may have been profitable, the manufacturer still saves money by having those car sales shift to other dealers.

None of this post answers the question of whether political bias was used in choosing which dealers to cut. However, it does make it clear that the choice to cut dealers was an economic, not political, one. You may recall Volvo and Subaru taking similar actions a few years ago.

Reliapundit said...



The Auto Prophet said...


You are right, it is small out of the whole dealer body, but it is a statistically significant sample. It takes a long time to do the study by hand.

Anonymous said...

What is your opinion of "The Mysterious Case of RLJ?"

Anonymous said...

Chrysler management had criteria to retain-terminate a dealer. The first step in the analysis process is for the Chrysler management and the car czar publish the criteria used to retain-terminate a dealer. The next step is verify the criteria was applied equally to all dealers. The third step, if required, is to challenge the criteria.

Anonymous said...

First, thanks for showing preliminary evidence that most dealerships probably give to the GOP - most are small businesses and this tends to be where they favor. And it is valid to argue that if the sample proportion echoes the population proportion, there is no bias.

However, your calculation is not statistically sound because your sample size is too low to get around margin of error. At n=25, theoretical MOE is around 31%. Yes, you can run a Z-test IF THE UNDERLYING DISTRIBUTION IS NORMAL.
But we don't know the underlying distribution of dealerships by donor choice in this example - it can be calculated, though.

So your post is a good start but actual evidence requires more in-depth legwork.

Chrysler said...

The analysis is quite impressive but there were many other factors apart from the donations and the criteria by chrysler to trminate contracts which worked against the dealership the most important of them being the slowdown in the economy which had a impact not only on the auto sector but also on the banking sector which was a major feeder for the auto industry banks failed to finance loans to consumers who earlier just made a phone call and a loan was approved.
Even consumers have become wary of having huge debts and treading cautiously.