Monday, May 01, 2006

Dow Chemical Follow-up

I seem to have attracted a lot of traffic based on my SELL rating on Dow Chemical.

My thanks to "sirlurksalot" for posting a link to a discussion on the Yahoo stock boards concerning my call. Reading the comments on my post was quite interesting. My favorite has to be the guy (jkohl2003) who just posted a link to an animated rally monkey.

Full disclosure on the people who posted: My logs show a lot of visits from people within the Dow Chemical Corporation, so many of those saying DOW should shoot up are probably employees and stockholders of the company.

Full disclosure on my side: I don't have any position in DOW (long or short) and wouldn't mind reversing my call if I can find data to support the position. Several of the good folks from Dow Chemical seem to have clicked on the Google advertising on this site, making them instantly my new favorite people. Lets see if further analysis can change my call to keep them happy.

One point made by "tobythelab" was that my Z-score based creditworthiness analysis disagrees with the S&P and Moody's credit ratings. First of all let me point out that I highlighted this very disagreement as well. Whenever I see a disagreement like this I eagerly anticipate a chance to learn something and I have been poring over the credit default literature since then. I found one particularly relevant paper from 2002: "Benchmarking the rating agencies: In search for the true credit rating transition matrix" By professors E. Altmann and H. Rijken of New York University and Amsterdam respectively. In this paper the study authors found that there was a high correlation between Z-scores and S&P/Moody's credit ratings but that the private agencies had a lag that averaged three years. In other words the Z-score is a predictor of the future S&P and Moodies ratings, as opposed to the opposite. I didn't know that before and it seems very useful, so I thank you for your challenge but I see nothing in that direction to change conclusions yet.

I can't find a publicly accessible copy of the PDF for you (my source is pay only), but this link should get you started. A lot of companies also have Nerac for papers.

Possibly the best reply came from "hottaire", who suggested that joint venture sales were not being taken properly into account for the sales ratios in the Z-score calculation. I found this to be very intriguing. "Total sales" are the so called "top line" income out of which are subtracted all the expenses and costs of good sold to arrive at a net income. The ratio of total sales to total assets is a strong component of the Z-score credit scoring system (a detail that shows hottaire has an eye for the fine points). If a company has a joint venture sometimes they record only the net income (the profit) from those ventures onto the balance sheet, in which case quite a few "sales" dollars would be missing from the computation. This was such a good question that it sent me back to the DOW 10-Q to pick apart their reporting. "Note I" to the report answers the matter by detailing numerous joint ventures and clarifying where the results are reported. Page 25 of the 10-Q then breaks out each segment (many of which represent one or more joint ventures) and shows the sales for each one. These were properly consolidated as gross sales to the income sheet. Thus, no sales were missed in the original calculations.

Finally just let me say, keep those comments coming. I eagerly look for chances to find out what I'm missing and who better than an employee of the corporation to point out strengths I may have missed. I also want to point out that my previous post, while it did give a "SELL" rating, had numerous positive comments about the company and it's stock and I wouldn't at all be surprised if it were up a bit in the next year. It just doesn't look cheap enough yet to be a highly attractive purchase to me. If you still hate me, feel free to take out your agressions by viciously clicking on those Ads you see (hah!).

Keep those comments coming. And hottaire and tobythelab are both official Wonks of the Day for their contribution.
Click back to the main page!

8 Comments:

Anonymous Anonymous said...

A comment on the JV contribution about which you already wrote. If I understand your conclusion from your analysis of the 10-Q, you concluded that indeed top-line sales from JV’s were being included. If that is correctly your conclusion then I believe you may have erred. The 10-Q reports “Equity in earnings of nonconsolidated affiliates by operating segment”. The key word here is earnings. A simple look at the numbers should be a good indicator that these cannot represent the sales. Looking back at the 10-K report, Dow reported “Dow’s share of earnings of nonconsolidated affiliates in 2005 was $964 million”. Think about Dow Corning as a gauge. I think hottaire had the right idea. The Z-score analysis done as you have shown here would account for all of Dow’s equity value in the assets, but only the earnings contribution and not the full value of the sales.

c_six_h_six

8:05 PM  
Anonymous Anonymous said...

After a bit of research and calculating I ended up with another question or two. If I have been reading correctly, the Z-score was formulated using annual income data. Do I have that correct? Because after running some calculations I get numbers well below 2 if I use quarterly results but above 3 if I use annual results. Do you get the same distressed indication if you use 2005 10-k data rather than 1Q06 10-q data?

A comment – by my calculations adjusting either assets or sales to approximate the JV impact on either is a small effect by comparison.

c_six_h_six

9:36 PM  
Blogger FinanceWonk said...

C6H6 (nice handle)-

I see what you are looking at but there are complexities here. The first issue is that this quarter they have started a new reporting method (see first paragraph of Note I in the 10-Q) so the 10-Q is actually differently reported from the 10-K. Dow Corning, for example, used to be in "Unallocated and other" but has now been moved to "Performance Chemicals" segment, which had 3.5 Billion in sales first quarter alone. Thanks to the new way of reporting the top chart on page 25 of the 10-Q seems to contain all the sales of the joint ventures. Comparing that to previous quarters shows that using the total revenues numbers for sales (which is what I did in the first place) does capture all the joint venture revenue.

9:51 PM  
Blogger FinanceWonk said...

C6H6:

One issue is that this quarter they have started a new reporting method (see first paragraph of Note I in the 10-Q) so the 10-Q is actually differently reported from the 10-K. Dow Corning, for example, used to be in "Unallocated and other" but has now been moved to "Performance Chemicals" segment, which had 3.5 Billion in sales first quarter alone. Thanks to the new way of reporting the top chart on page 25 of the 10-Q seems to contain all the sales of the joint ventures. Comparing that to previous quarters shows that using the total revenues numbers for sales (which is what I did in the first place) does capture all the joint venture revenue. I don't think I ever made it clear that I have been doing that correction from the very beginning.

How did you get a value over 3 using annual data? I used annual data from the start and got near 1. What number do you end up using for sales after adjusting for JVs?

As for Z-score over time and industry I actually have some data and charts on that, but didn't want to totally blow out my post with graphs and numbers. You are correct that DOW is not so bad "for the industry".

I would be interested in more thoughts on this, there may be a need for a third post and if we can find more sales or justify some industry differences I wouldn't mind changing my call and buying in. I'd rather be corrected and make money than save my ego and miss a chance

10:13 PM  
Anonymous Anonymous said...

We disagree on the reporting of the JV's.

Looking at he Income statement on page 3, I se Net Sales = $12,020 MM. Five lines down from that I see Equity in earnings of nonconsolidated affiliates = $168 MM. When I do the math between Net Sales and Income before Income Taxes and ... I add the 168.

12020 -9803-278-388-12+168+30+42-156 = 1623

Race ahead to Note I and the table on page 25. Net sales ($12,020 MM) are broken down the business segments (these are what are new - the segment designations). Then the second part of the table details the EBIT for each segment. The third part details the Equity earnings in ... As I read it the Equity earnings in the noncon affiliates is included in the EBIT calculations in the middle part of the table, but it is not included in the Net Sales. If it were, then they would be adding both the sales and the earnings to get to the Net Income before. Second, these are Dow’s share of the earnings in those affiliates based upon Dow’s equity share.

On my calc's, they should probably be ignored as I am not sure I understand the subtleties here. I believe the reported 3 was probably a screw up that I cannot recreate. But for the dominant X5 term (Sales/Totoal Assets) first I calc’d the Net Sales for the past 4 quarters as 46,648. Then I assumed that for the JV’s that Dow’s share of their Net Sales would be approximately 10x their earnings, or 1680 for the quarter. For the past 4 quarters that assumption yields 8,570 for the JV’s Net Sales. Adding the Net Sales from Dow and the Noncon affiliates I get 55,128. Total Assets I read as $45,564. Those two give me an X5 = 1.21.

c6h6

3:20 AM  
Anonymous Anonymous said...

With a bit more time than my hasty comment this morning allowed, I have reconstructed my analysis of Altman’s Z-score for Dow, which I will share here. For simplicity I have chosen to use YE2005 data from the recent 10-k report for two reasons: (1.) I need only one source so I don’t have to adjust for differences in 1Q05 and 1Q06 contributions, and (2.) I don’t see any significant changes in 1Q06 that would materially alter the result. Also for simplicity I used Altman’s “more convenient” specification as detailed on page 13 of his paper – Z= 1.2X1 + 1.4X2 + 3.3X3 + 0.6X4 + 1.0X5.


From the 2005 10-k I extracted the following data:
Net Sales 46,307
Current Assets 17,404
Current Liabilities 10,663
Long-term Debt 9,186
Other Non-current Liabilities 9,425
Total Assets 45,934
Investments in Noncon Affiliates 2,285
Retained Earnings 14,719
EBIT 6,963
Earnings from Noncon Affiliates 964
Shares Outstanding (Basic) 963
Share Price 44

From these I derived the following:
Working Capital 6,741
Total Liabilities 29,274
Market Value of EQ (MV) 42,207

Without any adjustment for the Nonconsolidated Affiliates (NCA’s) which is your position, I get the following:

X1 (WC/TA) 0.15
X2 (RA/TA) 0.32
X3 (EBIT/TA) 0.15
X4 (MV/TL) 1.44
X5 (S/TA) 1.01
Z 3.00

After some serious thinking I concluded that the most reasonable approach for considering the impact of the NCA’s is to back out their contribution to both EBIT and Total Assets. It seems unreasonable to adjust sales, etc. from the NCA’s but exclude their respective assets. So, following this logic I came up with:

Adjusted Total Assets (45,934 - 2,2850 = 43,649
Adjusted EBIT (6,963 – 964) = 5,999
Using these values I arrive at the following for a base Dow Z
X1 (WC/TA) 0.15
X2 (RA/TA) 0.34
X3 (EBIT/TA) 0.14
X4 (MV/TL) 1.44
X5 (S/TA) 1.06
Z 3.04

One problem that remains with the adjusted version is the Market Value of Equity (MV) term in X4. Surely Dow’s market equity valuation includes consideration of the NCA’s contribution to Dow’s earnings. The problem is how to adjust it. If we make the brazen assumption that MV would change proportionally to the EBIT change (i.e., MV will track the resultant earnings proportionally), then the MV would be 42,207(5999/6963) = 36,363. This would give an adjusted X4 term = 1.24 and a resultant Z-score = 2.92.

I hope this clarifies any confusion I may have created from my rather short response earlier.

c6h6

7:26 AM  
Blogger FinanceWonk said...

C6H6,

I'm glad you posted in such detail, now I see the source of much confusion. The Z-score is not the sum of the five X terms, it is a weighted sum and you appear to not be applying the weighting factors that apply to each term. I get very similar numbers for the X terms as you do, but the final Z-score is different because of the weighting terms.

The weighting terms are in the paper and also in the orignial post I put up.

Cheers, FW

9:26 AM  
Anonymous Anonymous said...

Maybe you didn't read my comment carefully. My Z score is dervied from weighted terms according to "Altman’s “more convenient” specification as detailed on page 13 of his paper – Z= 1.2X1 + 1.4X2 + 3.3X3 + 0.6X4 + 1.0X5". I noted this in my last comment (last sentence, first paragraph).

Perhaps did you use the simple ratios of the X terms and then mistakenly use the coefficients that are supposed to be applied for terms X1-X4 using the percentage values? Altman explains this importance of this as a Clarification on pages 12 & 13 of the paper you referenced. If I use the simple ratios as I calculated them but mistakenly apply the coefficients from the original equation, I get a Z-score of 1.21 for the Unadjusted case.

If you want to use coefficients of the original formulation then the ratios need to be the whole percentage values. Doing so gives the following for X1-X4:

X1 = 14.7
X2 = 32.0
X3 = 15.2
X4 =144.2
X5 remains as is (note the coefficient was simply rounded to 1.0).

Are we there yet?”

c6h6

10:06 AM  

Post a Comment

Links to this post:

Create a Link

<< Home

Some recommendations and links to friends Amazon Gift Certificates MEMS Consulting recommendation MEMS Discussion forum and information Nanotech Consulting