The following is a reprint of my January 8, 2008 RealMoney column.

In other articles, I have outlined the reasons why I think the semiconductor industry is poised for strong stock performance and why I think MEMC Electronic Materials (MEMC) is the best play on the sector.

But I also realize that a bullish semiconductor outlook right now involves making a grab at that falling knife. Therefore, I thought I should also let people know which semiconductor stock looks most vulnerable to a downturn.

I think that stock is Microsemi (MSCC).

Microsemi is a leading designer, manufacturer and marketer of high performance analog and mixed-signal integrated circuits and high-reliability semiconductors. Its products manage and control or regulate power, protect against transient voltage spikes and transmit, receive and amplify signals.

Microsemi has held up fairly well, handily beating the performance of the Semiconductor HOLDRs (SMH) over the last year. This may be due largely to its strong end markets, which include defense, commercial aerospace, industrial/semicap, medical, mobile connectivity and notebooks, monitors and LCD televisions.

More Questions Than Answers

To me, however, the strong end markets only raise questions concerning Microsemi’s fundamental performance. For example, with such strong end markets, why did its cash from operations fall by more than half in the year ended September 30, 2007, compared with the prior year? Why is its inventory rising faster than sales, and why is its gross margin slipping?

I turned to the company’s latest 10K in hope of finding answers.

To begin with, the area is highly competitive. According to the 10K (emphasis added), “some of our current major competitors are Freescale Semiconductor, Inc., National Semiconductor Corp. (NSM), Texas Instruments, Inc. (TXN - Annual Report), Koninklijke Philips Electronics (PHG), ON Semiconductor Corp. (ONNN), Fairchild Semiconductor International, Inc. (FCS), Micrel Incorporated (MCRL), International Rectifier Corp. (IRF), Semtech Corp. (SMTC), Linear Technology Corp. (LLTC), Maxim Integrated Products, Inc. (MXIM), Skyworks Solutions, Inc. (SWKS), Diodes, Inc. (DIOD - Annual Report), Vishay Intertechnology, Inc. (VSH), O2Micro International, Ltd. (OIIM) and Monolithic Power Systems, Inc. (MPWR).” Gosh, I wouldn’t want them to leave anyone out.

Yet competition is just the third risk factor among a list that runs more than 12 pages.

The company notes the decline in net income related to non-cash acquisition related charges, restructuring charges and other factors. Yet non-cash charges don’t quite explain the decline in cash flow from operating activity. Furthermore, with “non-recurring” charges being reported in each of the last three years I’m going to go out on a limb and say investors can probably expect more of them in the future.

A Questionable Acquisition

According to the 10K, the company completed a merger with PowerDsine on January 9, 2007 and subsequently renamed PowerDsine Ltd., Microsemi Corp. – Analog Mixed Signal Group, Ltd. (”AMSGL”). Later, it notes that it “provided a valuation allowance of approximately $9,534,000 as of September 30, 2007 on all of our net deferred tax assets related to AMSGL as we have determined that it was more likely than not that the deferred tax assets would not be realized.”

Deferred tax assets are realized when the company earns taxable income in future periods. I’m not a big fan of acquiring companies that will “more likely than not” fail to earn taxable income in the future. This was one of the contributors to the decline in cash flow.

Microsemi’s gross margin weakened in the latest quarter (see chart.)

Source: Zacks Research Wizard, compiled by William A. Trent

I think there is additional margin risk stemming from burgeoning inventory levels.

Source: Zacks Research Wizard, compiled by William A. Trent

Since a large percentage of costs at semiconductor companies is fixed, producing more units results in a lower cost per unit and higher profit margins. But many of the additional units Microsemi is producing are going into inventory rather than the hands of customers.

At some point, Microsemi is going to have to sell that inventory (by producing less than customers demand.) That will reverse the positive effect on future gross margins.

Valuation Too High

All this would matter less if the stock looked cheap. But on the basis of free cash flow yield, which is my favored metric, Microsemi looks more expensive than most of its peers.

Free cash flow in 2007 was less than $4 million. On an enterprise value of $1.56 billion, that amounts to a free cash flow yield of just 0.25%. The cash flow would have to grow 150-fold just to bring the yield on par with that of Treasury bonds.

Even using the company’s best cash flow on record ($36.5 million in 2006) the yield is just 2.35% – nearly a percentage point below that of Treasuries. If I thought the company could return to the 2006 cash flow level, then grow at the forecast rate, I would be willing to consider an investment.

But given the rising inventory, unprofitable acquisition and potential for further declines in gross margin, I won’t be holding my breath.

Disclosures: William Trent is long Semiconductor HOLDRS (SMH) and Maxim Integrated Products (MXIM). He holds put options against shares of Lam Research (LRCX).

William Trent currently owns put options against the shares of Lam Research (LRCX).

In a press release yesterday, the Semiconductor Industry Association said:

According to iSuppli, an independent market analysis firm, excess semiconductor inventories are increasing, reflecting expectations of growing demand in the second half of the year.

An increase in inventory may be justified if sales are expected to grow at the same pace or greater. This could be due to seasonal factors, a new product launch, or merely wishful thinking on the part of the company building the inventory. I’m continuing my series on semiconductor inventory trends with a look at the sequential change in inventory between the last reported quarter and the previously reported quarter, and comparing that figure to the change in expected sales between the current quarter and the next quarter. I used Zacks Research Wizard to collect the data.

The metrics I used are imperfect at best, as the change in inventory in any two quarters will not necessarily correspond exactly to the expected sales growth over the subsequent two quarters. However, it should be useful as an indication of whether sales levels generally are expected to grow sufficiently to absorb the inventory.

Of 47 companies that passed the screen, I found that 20 of them have inventory growth within five percentage points of expected sales growth, which seems reasonable. I also found that there were more companies that have expected sales growth greater than past inventory growth than there were the opposite, but I would caution that semiconductor sales estimates have been coming down recently and the current data may be too optimistic.

The five companies for which inventory either declined or grew at a slower rate than expected sales growth by the widest margin were Silicon Labs SLAB, Silicon Image (SIMG), MEMC Electronic Materials (WFR), Applied Micro Circuits (AMCC) and Stats Chipp (STTS). In the case of Silcon Labs the change may be due in large part to the recent sale of its handset chip division, which accounted for half the company’s sales.

The five companies that had the largest increase in inventory, relative to expected sales growth, were Actions (ACTS), Cypress (CY), Monolithic Power (MPWR), Atheros (ATHR) and Linear Technology (LLTC).

Disclosure: William Trent has a long position in SMH.

Update: The original post contained a data error.
In the interest of digging deeper into the semiconductor oversupply issues, this post continues a series of data gathering on important ratios for companies in the industry. Hopefully the process will provide insight toward the companies better (or worse) positioned to take advantage of the next upturn or weather the downturn.

Yesterday I used Zacks Research Wizard to get the recent Cost of Goods Sold (COGS) and Inventory levels for semiconductor industry participants over the last several quarters. I made some modest limitations on the share volume and market cap, but still ended up with more than 50 names. I used trailing twelve month COGS and the average of the last five quarters (for a beginning, ending and average) of inventory to calculate Days Sales in Inventory.

Higher inventory levels relative to sales indicates a greater likelihood that the company will need to reduce prices, reduce production or take a write-off, all of which would reduce gross profit margin. In this post I compare the current DSI to the DSI in the same quarter one year ago. This should mitigate any seasonal effects, such as ramping inventory ahead of holiday sales, that might distort sequential comparisons.

The companies with the biggest increase in DSI may have the most trouble in the event of an industry downdraft. Even if semiconductor sales remain strong they will need demand to catch up with their current capacity and may not see as much benefit as other manufacturers. The five companies with the largest year/year DSI increase are Silicon Labs SLAB, Applied Micro (AMCC), Zoran (ZRAN), Monolithic Power (MPWR) and Triquint (TQNT).

The companies with the greatest reduction in DSI, by contrast, may be poised for margin expansion as they replenish inventory levels and ramp up production to meet demand. The five companies with the biggest decrease in DSI are Large Cap Watch List (Track at Marketocracy) member MEMC Electronics (WFR), Advanced Micro Devices (AMD - Annual Report), Atheros (ATHR), Conexant (CNXT) and Intersil (ISIL).

The complete list follows.

DSI As Percentage of DSI One Year Ago:

Disclosure: William Trent has a long position in SMH.