Tower Semiconductor: The Books Look Clean

| About: Tower Semiconductor (TSEM)


The value of TPSCo as recorded on the balance sheets of Tower and Panasonic should be different. This difference is not evidence of accounting fraud.

TPSCo should be recorded on Tower's balance sheet at fair market value as required by purchase price accounting under US GAAP.

Panasonic must record TPSCo at historical cost plus cost of improvements minus the accumulated depreciation of those assets.

Comparing the estimated useful life of equipment at a foundry specializing in analog ICs to one that focuses on digital ICs as Ben Axler does is misguided.

There is no reason to believe Axler's analysis of Tower Semiconductor should warrant additional scrutiny of Tower's books by the SEC.

Last week, Ben Axler of Spruce Point Capital Management published an article on Seeking Alpha making a case to short Tower Semiconductor (NASDAQ:TSEM). He also published a more detailed 79-page analysis on the fund's website. While I strongly disagree with large chunks of the analysis, there are a few interesting points presented. Page 74 of his presentation is titled It's Unanimous, Wall Street Analysts Say "Buy!" Whenever one side of a trade is significantly more popular among the investment community, this is a good opportunity to start analyzing the stock in hopes of finding evidence to create a strong contrarian thesis. When making a short case for a stock that is unanimously loved by Wall Street analysts, you will almost certainly be vilified by those long the stock. I applaud Ben Axler's efforts on doing his own homework and his willingness to make contrarian picks. However, in the case of Tower, I believe he is wrong.

Accounting of TPSCo Is Correct

The most brazen claim in Ben Axler's analysis of Tower is his assertion that Tower Semiconductor is materially misrepresenting the value of TPSCo's assets. TPSCo in a joint venture between Tower and Panasonic that was created in 2014. This joint venture was created by Panasonic moving three fabs into a separate entity and selling a 51% stake in this new entity to Tower. Axler claims that Tower "inflated the value of the JV's assets from approximately $100m to $360m in order to book a $166m bargain purchase gain to bolster its equity." He then uses Tower's Q2 2014 Consolidated Financial Statements and a Panasonic Press Release from December 20, 2013 to support this claim.

When analyzing the economic value of the TPSCo JV, it would be reasonable to expect that Tower and Panasonic would give identical numbers when looking at the total assets and net assets of TPSCo. Because the numbers being reported by Panasonic and Tower are vastly different, Axler believes that Tower is employing some accounting shenanigans. On page 33 of his presentation, he states, "What could be greater evidence of a potential accounting scheme than 2 companies marking an identical transaction at different values! TSEM's valuation is 250%+ greater than Panasonic's!" If he were a little more careful, he would have noticed that Panasonic's presentation describing the net assets and total assets of TPSCo on April 1, 2014 was made over 100 days before this date. Tower's presentation of TPSCo's total and net assets as of March 31, 2014 was made well after this date. While the timing difference may cause some differences, it certainly does not fully explain the large gap between the two presentations.

Click to enlarge

(The above graphic is taken from Axler's article. These are screenshots of filings from Tower Semiconductor and Panasonic.)

As a result of purchase price accounting (under US GAAP), Tower must record and present the FMV (fair market value) of TPSCo on its balance sheet. Tower acquired a 51% stake in the joint venture by granting Panasonic 870,454 shares of TSEM stock. As of the date of the completion of this transaction, the value of those shares at the close on March 31, 2014 was approximately $7.96 million. The fair market value of the total assets of TPSCo as of the transaction date was recorded as $361,212,000 and the net assets was recorded at $180,935,000 on Tower's financial statements. An independent third party made these valuation estimates and a member firm of Deloitte Touche Tohmatsu Limited audited these estimates. In Axler's presentation, he suggest that this member firm of Deloitte is not an auditor that he trusts.

The best way for a large publicly trading company to ensure investors that its financial statements are accurate and free of any material misrepresentations is for the company to hire one of the Big Four accounting firms as an auditor. The Big Four audit firms are PwC, Deloitte, EY and KPMG. While some may favor one of these firms over the others, the truth is that all four firms are very similar. Here is a quick history of how the Big Five became the Big Four. Arthur Andersen was the international professional services company that fell apart in 2002 that was a member of the Big Five. The event that sparked the collapse of this once mighty accounting firm was its involvement in the Enron scandal in which auditor Arthur Andersen was found to be complicit in the accounting fraud scheme of the former energy giant. Given the size of Tower and the relatively small fees that the Israeli company pays to Deloitte, it seems highly unlikely that this Big Four accounting firm would jeopardize the integrity of the company and its annual revenues of well over $20 billion for an audit fee of less than $1 million from Tower Semiconductor as documented on page 99 of the company's most recent 20-F filing. Axler points out Tower's auditor, Deloitte Brightman Almagor, Zohar & Co., has a small client base. He claims Tower's use of this firm as an auditor as a reason to be suspicious of Tower's accounting. Of course, he doesn't really consider how competitive the Israeli accounting market is. On top of the intense competition from other Big Four accounting firms, Israel has one of the highest per capita number of accountants in the world. Intense competition as an auditor in Israel also comes from firms outside of the Big Four. What would he propose as an alternative to using Deloitte? Are the other members of the Big Four really significantly better than Deloitte? I highly doubt it. Had Tower Semiconductor elected to use an auditor outside of the Big Four, I suspect his argument about Tower's accounting problems would shift to "Why would a semiconductor company with a billion dollars of annual revenue voluntarily choose to use a small accounting firm as an auditor if it has nothing to hide?"

On top of reviewing the estimates of the fair market value of TPSCo's assets from the independent third party and Deloitte, anyone considering an investment in Tower Semiconductor should also perform at least a "sanity check" to make sure those reported numbers make sense. TPSCo's revenues from Panasonic are expected to be between $360 million and $420 million per year. Almost half of the wafer capacity of TPSCo is still lying idle. By the time Tower fills the remaining capacity with sales to companies other than Panasonic, annual revenues in excess of $700 million are expected. This is what we should expect by 2018. One thing to note about Tower's foundry business (which is true at virtually all other foundries as well) is that it has high fixed costs and low variable costs. What this means is that incremental gross margins from revenue gained by using underutilized fab capacity is very high. I estimate the incremental gross margin on the annual $300 million of third party revenue at TPSCo (by 2018) to be in the range of 60-80%. That means TPSCo could add an additional $180 million - $240 million of gross profit at TPSCo by 2018. This comes on top of the sales and profits from recurring business from Panasonic. With this information, it is easy to see that the $361.2 million estimate of the fair market value of TPSCo's total assets as of March 31, 2014 is not ludicrously high. From this perspective, this estimate seems low.

The main reason for the discrepancy between the way Tower and Panasonic recorded the TPSCo transaction is relatively simple to explain. Tower must record the total assets and net assets of TPSCo at fair market value. Panasonic must record the total assets and net assets of TPSco under a different method. Panasonic must report the historical cost of buying/building the fabs plus the cost of improvements over the years minus the accumulated depreciation of those assets. The odds of that formula creating the exact same result as a fair market value estimate by an independent third party is extremely small. Not only does the difference in the reporting of the transaction by Panasonic and Tower not indicate accounting problems, but it would be very suspicious if the two reported the same total and net assets for TPSCo on their respective balance sheets. One question that you should ask yourself when trying to detect fraudulent accounting is to ask yourself whether a company attempting to mislead investors by cooking the books would choose a method that is so easily detectable. By doing this, you would have realized that the difference in the reporting of TPSCo's assets by Tower and Panasonic is nothing more than standard accounting (required by law) that frequently produces balance sheets that do not reflect economic value.

Russell Ellwanger's Education And Experience

On page 23 of Spruce Point Capital Management's presentation, it states "In the semiconductor manufacturing industry, it is typical for the CEO/President to have an advanced educational degree." Certainly, that is true. However, it is possible for a CEO to lead a company to great success in the technology industry without a degree at all. You may have heard of a couple guys by the names of Bill Gates, Larry Ellison, Michael Dell and Mark Zuckerberg. Of course, these are some of the greatest minds of our era and are exceptions to the general rules of the value of a college education. While there are many people who can write code and understand consumer electronics without ever stepping foot in a college classroom, it is much more difficult to find someone who has a great grasp of semiconductor design and manufacturing without some rather impressive academic credentials. However, the lack of an advanced degree in engineering or the sciences does not make it impossible for someone to lead a great semiconductor company.

Because Russell Ellwanger, CEO of Tower Semiconductor since 2005, does not mention his academic credentials in his official biography like the other executives at the company, Ben Axler implies that Ellwanger has never attended a university. While I tried looking up information about his academic history online, I was unable to find anything. However, if someone were interested in his academic credentials as Ben Axler is, all he has to do is simply ask Russell Ellwanger or contact the investor relations team at Tower Semiconductor. If Axler had simply asked Noit Levy-Karoubi (Vice President of Investor Relations and Corporate Communications at TowerJazz) or Kenny Green (CCG Investor Relations - US), he could have easily found out the truth about Ellwanger's academic history. Their contact information is listed on the TowerJazz website here. From my experience interacting with them, they are both actively answering questions from investors and respond in a much more timely manner than the vast majority of companies in the market today. By the way, Axler's assumption that Ellwanger never attended a university is false. I know this because I took the effort to ask. When you make a short thesis centered around the idea that a company has fraudulent financials, it's probably a good idea to contact the company and ask them questions about the financials. Despite the fact that Axler is likely not the favorite person among Tower's management right now, I strongly suspect they will be more than willing to field questions from him.

Axler is wrong in implying that Russell Ellwanger is not qualified to be the CEO of a semiconductor company that will soon have over a billion dollars of annual revenue because he does not have an advanced engineering degree. While looking at his qualifications, it would be an enormous mistake to only analyze his accomplishments from his 20s and ignoring his over 35 years of experience in the semiconductor industry. According to his biography on the TowerJazz website, Ellwanger began working at Philips Semiconductor in 1980 and has served as an executive at other well-known semiconductor companies including Novellus Systems and Applied Materials. Axler does not see the possibility that Ellwanger is a great semiconductor executive. This implicitly states that those who promoted Ellwanger through the ranks over decades at Philips Semiconductor, Novellus Systems and Applied Materials are also incompetent to let someone who is so "unqualified" to reach high level management positions at these firms. It would also imply that Tower Semiconductor's board, which consists of people with advanced degrees in business and engineering, did not do their due diligence before appointing Russell Ellwanger as CEO.

Despite the facts stated above about Tower Semiconductor's CEO's qualifications, Ben Axler believes he can analyze Ellwanger's qualifications to be the CEO of Tower better than the management at Tower Semiconductor, Applied Materials, Novellus Systems and Philips Semiconductor. Based on the 79 page presentation about Tower Semiconductor, it is quite obvious that Axler, unlike Ellwanger, does not have a background in engineering. Anyone with even a basic understanding of the foundry business can explain why the estimated useful life of fab equipment at a foundry specializing in RF and other analog ICs is longer than that of fab equipment used in the construction of digital ICs. A discussion about the business itself is completely omitted from the presentation.

Estimated Useful Life of Foundry Equipment

I posted the below comments on Axler's original article about Tower Semiconductor on this website. On page 57 of Axler's presentation that was posted on Spruce Point Capital Management's website, there is a chart that shows "Accounting Assumptions for Depreciation Out of Line With Peers." He correctly identifies that TowerJazz assumes the estimated useful life of its equipment is 15 years. He then goes on to state United Microelectronics (NYSE:UMC) assumes 3-11 years, Semiconductor Manufacturing International assumes 5-10 years, Vanguard International Semiconductor assumes 3-5 years and Taiwan Semiconductor (NYSE:TSM) assumes 2-5 years. All of this is true.

I want to point out the glaring mistake of comparing the useful lives of TowerJazz's equipment and that of Taiwan Semiconductor Manufacturing. (The same argument applies between TowerJazz and the other companies he lists as well.) The useful life of equipment used for digital ICs is short because the cost of manufacturing comes down after an expensive node shrink (that requires lots of very expensive brand new lithography equipment that only gets more expensive with each generation.) The limit of when cost benefits disappear for advancing the lithography process in digital ICs has not yet been reached. When it comes to analog ICs (TowerJazz's domain), there are numerous RF ICs that gain little to no cost benefit as a result of a node shrink. In fact, costs will start to rise as a result of a node shrink in some applications. As a result, analog foundry equipment should have a significantly longer life than that of its digital counterparts.

A fab equipment's life is rarely dependent upon how many wafers can be processed until the machinery breaks down. The life of lithography equipment is dictated by the availability and cost benefit of more advanced lithography equipment. Because Tower (and its analog competitors) cannot achieve cost savings with new equipment in many situations, the Israeli company and its peers will likely use some of its current fab equipment close to indefinitely. A 15 year life for a very large chunk of Tower's equipment is more likely to be too conservative of an estimate instead of wildly aggressive as Axler believes.

One way for a firm to manipulate its financial statements is to change its estimates of the useful life of its PP&E. By increasing this estimated number by a couple years, annual depreciation expenses would be lowered. As a result, bottom line net income would get increased. Whenever a company changes its estimated useful life of its equipment or discloses significant changes in its depreciation policy, an investor should be wary of the possibility of accounting shenanigans. Investors should be even more cautious if the company is involved in a business that traditionally requires enormous capital expenditures (like the foundry business). Ben Axler appears to be doing prudent due diligence here.

This statement in Note D of the Q2 2015 financial statements is a bit scary - "it was determined that the estimated useful lives of machinery and equipment should be extended to 15 years from 7 years." However, I cannot see a reasonable basis for expecting the estimated useful life of Tower's analog fab equipment to only be 7 years. The correct analysis in this case is there was an error in estimating the useful life of the equipment and this error caused a material misrepresentation of earnings. However, unlike Ben Axler, I see that TowerJazz UNDERSTATED its true earnings in the past as a result of the use of a bad estimate of the useful life of its equipment. It is likely that earnings will still be underrepresented by using a 15 year estimated life. It is absurd to believe that the estimated useful life of Tower's equipment should be anywhere close to that of the foundries focusing on digital ICs that he compares Tower against in his presentation.

TSEM's Larger Shareholder Has Not Sold Any Shares Since The Turn Of The Century

Page 68 of Spruce Point Capital Management's presentation is titled TSEM's Largest Shareholder "Long-Term" View. This page states that Israel Corporation stated it "has no intention to trade or sell the ordinary shares [of Tower Semiconductor] and will hold them as a long-term strategic investment." Axler states on the same page that Israel Corp "Completely liquidate[d]" its 39% position. I will just copy an excerpt of Lisa Thompson's comment on Axler's article about Tower Semiconductor. Lisa states, "With regard to Israel Corp and Kenon specifically, they have never sold a single share. After Kenon's split off from Israel Corp, it is their stated policy to wind down the company and issue shares to their shareholders. The major shareholder of Israel Corp, and therefore Kenon, and now Tower, is Idan Ofer, which still holds his same position in Tower and has not sold a single share since 1999." Here is a Schedule 13-D that explains the transactions that occurred. It is misleading to claim that "TSEM's largest shareholder" has liquidated his entire position when his position has essentially been shifted from the shareholder's left pocket to his right pocket.


I believe Ben Axler and Spruce Point Capital Management are incorrectly analyzing Tower Semiconductor's accounting practices and future business prospects. Here is a quick overview of Tower Semiconductor's unique business model in the foundry business. The two most egregious claims are discussed above relating to the discrepancy in the accounting of TPSCo between the JV partners and the change in the estimated useful lives of Tower Semiconductor's equipment. There is no reason for the SEC to take another look at Tower's books as a result of Axler's presentation. On the points I discuss above, I believe Axler is wrong. He has made some serious allegations against Tower's management that are just not true. (There are numerous other problems with his presentation that I did not discuss in this article.)

Disclosure: I am/we are long TSEM.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.