Heavy Lifting - thoughts and web finds by an economist
     I also contribute to Division of Labour Load HL's Front Page
Friday, May 15, 2009

Any Pattern in Chrysler Dealership Closings?

It took some work, but I was finally able to get the list of Chrysler dealerships on the chopping block into MS Excel and eventually accumulate the number of dealerships per state. There might be another tally by state but I couldn't find one via Google.

Here's a scatter plot of the number of dealerships per one million population against state economic freedom and whether the state went for Obama in the 2008 election:



Here is a tabulation by state of the number of dealerships to be closed.

+------------------+
| state numdea~s |
|------------------|
1. | AK 1 |
2. | AL 11 |
3. | AR 8 |
4. | AZ 5 |
5. | CA 32 |
|------------------|
6. | CO 12 |
7. | CT 7 |
8. | DE 3 |
9. | FL 34 |
10. | GA 13 |
|------------------|
11. | HI 1 |
12. | IA 22 |
13. | ID 3 |
14. | IL 44 |
15. | IN 22 |
|------------------|
16. | KS 16 |
17. | KY 9 |
18. | LA 17 |
19. | MA 12 |
20. | MD 17 |
|------------------|
21. | ME 4 |
22. | MI 40 |
23. | MN 19 |
24. | MO 27 |
25. | MS 6 |
|------------------|
26. | MT 4 |
27. | NC 14 |
28. | ND 8 |
29. | NE 8 |
30. | NH 6 |
|------------------|
31. | NJ 30 |
32. | NM 4 |
33. | NV 5 |
34. | NY 28 |
35. | OH 47 |
|------------------|
36. | OK 12 |
37. | OR 9 |
38. | PA 53 |
39. | RI 1 |
40. | SC 11 |
|------------------|
41. | SD 7 |
42. | TN 14 |
43. | TX 50 |
44. | UT 10 |
45. | VA 26 |
|------------------|
46. | VT 2 |
47. | WA 15 |
48. | WI 18 |
49. | WV 17 |
50. | WY 5 |
|------------------|
51. | DC 0 |

Here are the results from a negative binomial estimation where the number of dealers is the dependent variables:


. nbreg numdealers obama econfreedom pcinc,r

Fitting Poisson model:

Iteration 0: log pseudolikelihood = -336.534
Iteration 1: log pseudolikelihood = -336.534

Fitting constant-only model:

Iteration 0: log pseudolikelihood = -189.48901
Iteration 1: log pseudolikelihood = -187.24421
Iteration 2: log pseudolikelihood = -187.22283
Iteration 3: log pseudolikelihood = -187.22283

Fitting full model:

Iteration 0: log pseudolikelihood = -184.44899
Iteration 1: log pseudolikelihood = -184.10457
Iteration 2: log pseudolikelihood = -184.0962
Iteration 3: log pseudolikelihood = -184.09619

Negative binomial regression Number of obs = 50
Dispersion = mean Wald chi2(3) = 6.87
Log pseudolikelihood = -184.09619 Prob > chi2 = 0.0763

------------------------------------------------------------------------------
| Robust
numdealers | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
obama | .4999501 .2993714 1.67 0.095 -.086807 1.086707
econfreedom | .3463842 .2685567 1.29 0.197 -.1799772 .8727456
pcinc | -4.89e-06 .0000266 -0.18 0.854 -.000057 .0000472
_cons | .2338063 2.017646 0.12 0.908 -3.720708 4.188321
-------------+----------------------------------------------------------------
/lnalpha | -.5879579 .1928054 -.9658497 -.2100662
-------------+----------------------------------------------------------------
alpha | .5554604 .1070958 .3806596 .8105306


There isn't much here. There is weak evidence that states that voted for Obama in the presidential election of 2008 suffer more closed dealerships. Perhaps that is a political calculation? If these states suffere the most but are also the safest in terms of Obama's future presidential hopes then perhaps the number of dealership closings is non-random. However, economic freedom and per-capita income do not seem to have any relationship to the number of dealerships closing.

Stata data file

Labels: , , ,

Comments:
It seems more likely that those states with the largest number of dealerships in danger of being closed would be the most likely to vote for Obama in the first place in order to increase the chance that they would be saved. It's possible that those states didn't have as many closures as they would have under McCain, but of course there's no way to test for that.

Your results might just indicate that the states with the largest number of at-risk dealerships before the election were more likely to vote for Obama. But these states still had the most dealership closures after the election.
 
Not sure I trust p values for a limiting distribution with 50 obs and small sample sampling distribution unknown, but did you think maybe it's just selection bias? Swing states that were slumping seemed to have swung Obama's way, and are most likely to have ANY business close, including (maybe especially) auto dealerships.
 
Post a Comment



Purchase




Le Chai - galerie du vin



Support



Popularity

Posts that contain Craig Depken per day for the last 90 days.



Contacts

Heavy Lifting's Main Page
Email Me
Atom Feed

WWW
Heavy Lifting




Great Links

Money I Found Today

Heavy Lifting - Firehose style (56k warning)



Recent Posts

- Any Pattern in Chrysler Dealership Closings?




Archives


home


Visitors
Site Meter Blogroll Me!



Credits

Modified maystar design
powered by blogger