Top Analyst - Research Note

[Link "Bezugnahme auf unsere Vorgängermeldung über Mark Edelstone." auf 62.146.24.165/... nicht mehr verfügbar]

Autor: Mark Edelstone

Originalexzerpt von Morgan Staley Dean Witter

The Semiconductor Industry
Semiconductor Fundamentals Are Weak but Fed Action Should Begin to Help

Due to significant weakness in demand in most of the key end markets, including PCs, servers, storage, wireline and wireless communications, and digital consumer applications, a record number of around five dozen U.S. semiconductor companies preannounced negative first-quarter earnings surprises. Based on the data, it appears that the global semiconductor industry will decline by around 20% sequentially in the first quarter, making it the worst sequential decline since the first quarter of 1985. With the first half of this year off to a tough start, it is clear that 2001 will produce poor results, and the magnitude of the recovery in 2002 will be predicated on a reacceleration in economic growth and a sharp reduction in capital spending by the global semiconductor manufacturing companies.

Revenue Growth has Continued to Decelerate Sharply and Turned Negative In March

From an end market perspective, PC demand was disastrous in the fourth quarter, with 5% sequential and 4% year-over-year unit growth, and demand has remained soft thus far in 2001. Wireless handset demand came in at the lower end of expectations in 2000, and wireline communications demand was pressured by a slowdown in carrier spending and financing problems for broadband deployment. Thus far in 2001, we have seen earnings warnings within the enterprise networking, optical networking, wireless, PC, server and storage markets. Given attempts by leading OEMs, EMS companies and distributors to build buffer inventories when semiconductor supply constraints were in full force during the 1999/2000 timeframe, the slowdown in end demand has caused virtually every company in the worldwide semiconductor industry to be adversely impacted by an inventory correction.

On a year-over-year basis, the semiconductor industry's growth rate peaked at 52% in August 2000 and we believe that year-over-year growth turned negative in March (Exhibit 1). Based on recent order trends, it appears that second-quarter revenues will likely decline over the first-quarter level unless turns business (orders that are received and shipped for revenue in the same quarter) improves. Although the absolute level of semiconductor industry revenues should begin to stabilize sometime in the second quarter, tough comparisons can be expected to cause year-over-year growth rates to decelerate sharply until August or September. We currently expect negative comparisons to be recorded in each of the last 10 months of 2001, and the peak declines will likely be negative 20% or worse in the third quarter. Following 37% growth in 2000, we currently expect the industry to decline by at least 10% this year, and recent order trends suggest that a 15%-20% decline is possible.

The Weak Economy Has Pressured Demand...

Due to the ability to constantly innovate and provide ICs that are smaller, cheaper, faster and provide enhanced functionality over prior generation devices, the semiconductor industry is characterized as a high-growth, value-added sector. During the last 40 years or so, the global semiconductor industry has enjoyed compound annual revenue growth of 17%, which is roughly five times faster than average annual global GDP growth of 3.6%. However, due to the capital intensity, commodity orientation and fixed-cost nature of the industry, it is highly cyclical. In general, we believe that the cyclicality is driven by the capital intensity of the industry and the fact that approximately two-thirds of the industry's revenues are derived from products that are supplied by multiple competitors. When combined by the fact that most management teams tend to increase capital spending when margins expand and reduce spending levels when margins contract, the industry has been highly prone to transitioning between periods of excess capacity and capacity constraints, and this drives the industry's underlying cyclicality.

Following a snapback to normalcy in 1999, where the global economy and the worldwide semiconductor industry grew close to trend-line rates of 3.6% and 19%, respectively, explosive economic growth (4.9%) and capacity constraints led to semiconductor industry revenue growth that more than doubled the trend-line rate in 2000. In hindsight, it appears that pent-up demand from Y2K issues and the collapse of dot-com spending led to an unsustainable bubble of IT spending in the first half of 2000. With the completion of these spending binges, financing pressures in Internet-centric growth, and a sharp deceleration in global economic growth, near-term demand within the computer, communications and consumer electronics markets started to come under significant pressure in the second half of 2000. Despite recent actions from the Fed, the global economy is currently in the midst of one of the sharpest economic slowdowns on record, and it can be expected to continue to pressure overall demand in all of the key end markets.

...And Weak Demand Has Promoted a Broad-Based Inventory Correction

We estimate that inventories at the major computer, communications, EMS companies and distributors declined 5% sequentially in the fourth quarter, which we believe is the smallest fourth-quarter decline on record (Exhibit 2). The average decline in the prior three years was 16%. Inventories increased nearly 20% over the fourth quarter of 1999 and are about 30% higher than the secular low reached in the fourth quarter of 1998. As demand softened and semiconductor consumers attempted to reduce their own inventories, semiconductor company inventories increased. Two-thirds of the companies in our universe saw inventories increase sequentially in the fourth quarter, and the average increase for all of our companies was 16%. Inventory levels will likely increase further in the first quarter, which will clearly represent a bad trend for the foundries and semiconductor companies with fabs. Given the fixed-cost nature of the semiconductor industry, meaningful declines in capacity utilization will pressure margins as overhead levels are under absorbed.

The Risk of Negative EPS Surprises Remains High but Fabless Companies Should Fare Better than Those with Fabs

Due to the sharp deceleration in growth caused by the economic weakness and the concomitant inventory correction, most semiconductor companies currently have negative book-to-bill ratios, and declining backlogs and an increasing dependency on turns business limits near-term visibility and suggests that negative earnings surprises should remain high. As a general rule of thumb, the earnings risk for fabless companies should be less than those companies with their own manufacturing facilities. The manufacturing intensive companies will be incrementally hurt by under absorbed manufacturing overhead, while companies with commodity-oriented products will be hurt by a lack of pricing power.

Although they can still be expected to be adversely impacted, fabless companies with proprietary products will fare much better in the current environment. Given their variable cost nature, the EPS impact from lower-than-expected revenues will be significantly less for the fabless companies. For example, we estimate that EPS results would likely decline in line to 50% more than the any incremental decline in revenues for most of the fabless companies. In contrast, the gross margin pressure faced by most of the manufacturing intensive companies would likely cause earnings to decline by 2-3 times as much as any decline in revenues.

The Semiconductor Stocks Remain in a Tug of War Between Lower Interest Rates and Weaker Fundamentals...

Following a sharp advance in January, the SOX index has declined 21% since the Fed cut rates on January 31. However, we believe that the semiconductor capital equipment stocks have helped to support the SOX index, as the average semiconductor stock in our universe has declined 28% since January 31. While the SOX index is up 0.2% this year, our universe of stocks has declined 5%, and this compares to an 11% decline in the S&P 500, and a 21% decline in the NASDAQ Composite index. Although we do not expect the fundamental environment for the semiconductor sector to improve until the fourth quarter, we believe that the average stock has already discounted most of the near-term negative news. However, we believe that a sustainable advance is unlikely to occur until year-over-year revenue growth has reached the trough in the third quarter.

...But Fed Intervention for the Third Time has Historically Been a Charm for Most Semiconductor Stocks

Year-to-date, we believe that the semiconductor stocks are behaving right in line with normal historical parameters. Due to the sharp sell-off last year, we have expected the SOX index to be locked in a volatile trading range during the first 6-9 months of this year, with limited opportunities for a sustainable advance until the industry's year-over-year growth rate reaches a trough in August or September. Furthermore, we believe that the stocks will likely conform to normal trading patterns before and after interest rate cuts engineered by the Fed. Since very little of the Fed's surprise interest rate cut on January 3 was discounted by the stocks, the Fed action promoted a sharp advance in most semiconductor stocks during January. However, due to the discounting factor and the fact that investors began to realize how bad the deterioration in fundamentals was likely to be, the average semiconductor stock peaked when the Fed cut rates for the second time on January 31.

The Fed has supported six distinct periods of monetary ease since the early 1980s, and the semiconductor stocks tend to perform quite well in the 3, 9, and 12-month periods following the third interest rate cut (Exhibit 3). On a 12-month basis, the average semiconductor stock increased in absolute and relative terms (versus the S&P 500) in all six periods, with an average absolute gain of 92%, which was 65 percentage points better than the S&P 500. Unlike the mixed performance that has been encountered in the three-month period after the first and second interest rate cuts, the semiconductor stocks enjoyed absolute and relative advances in five of the six periods since the early 1980s after the third cut in interest rates. During the three-month period following the third interest rate cut, the stocks recorded an average gain of 18%, which was 10 percentage points better than the S&P 500.

The six-month results are a bit more mixed and support our view for a choppy trading range environment during the next one to two quarters, as the stocks contend with deteriorating near-term fundamentals, expectations for a stronger economic environment next year, and the sharp decline that has already impacted every semiconductor stock since last summer. While the next three to six months is likely to remain challenging for investors, we believe it should represent a solid opportunity to build positions in the stocks of many of the top-tier semiconductor companies. We continue to believe that the fabless companies with proprietary products and strong product cycles will have the least amount of additional earnings risk. With their variable cost business models, the fabless companies should be able to minimize any overhead absorption problems and the fabless companies with proprietary products should retain pricing power and benefit from lower wafer and assembly/test costs as industry utilization rates decline.