ML analyst reads CFZ:
No Hurry: A top analyst is bullish on tech, but isn't eager to jump in now
By Eric Savitz in Barrons
An Interview With Steven Milunovich -- Tech stocks are
back. While still far below their bubble-era peaks, chip,
networking and software shares have all lately notched fat gains.
While fundamentals remain grim, there's a growing belief that
the vicious technology recession may have bottomed. So far,
however, the proof is sparse. Are investors correctly anticipating
the turn? Or are tech stocks set up for another slide? To assess
the situation, we tapped Steven Milunovich, global technology
strategist at Merrill Lynch. A former hardware analyst,
Milunovich has been cautious on the technology sector for
months now. In the Q&A below, he provides some safe havens
-- and warns we may be seeing a new bubble.
Eric J. Savitz
Q: Steve, technology fundamentals seem dicey at best -- and yet
investors have been flocking back.
A: There's still a bias to own tech. At the first sign of better
fundamentals, the stocks rally. You could argue that it means
we're never going to get down to the lower valuations many of
us would like. Or you could argue that we haven't seen the final
capitulation. I'm in the latter camp.
Q: Why don't you believe?
A: I view this rally like the one we had in April: Tech is in a bear
market, and this is a bear-market rally.
Q: But there have been a few hopeful signs, particularly in
chips.
A: True. And historically, when semiconductor shipments start
to get less worse, it's been a buy signal. Also, the tech profit
cycle likely troughed in the third quarter, and will begin to
improve from here. That also tends to be a buy signal. Four out
of five times, playing this kind of second derivative game gives
the right call. But this is the one out of five times when it
doesn't. We need time to work off the excesses of the bubble.
Technology evolves in waves of about 15 years and, right now,
we're transitioning from the second wave, which was
PC-centric, to the third, which is network- and Internet-centric.
In these transitions, tech stocks tend to underperform.
Q: This has happened before?
A: We had a transition like this in the mid-'Eighties, when we
moved from the first wave, which was mainframe-centric, to
PCs. As you go into a new wave, you initially have a euphoric
period. The Internet bubble years were a lot like 1982 and 1983,
when you had a PC and disk-drive boom.
Q: And booms tend to be followed by busts?
A: Right. One thing that happens is that the leading companies of
the previous wave go into secular decline. The next wave tends
to be lead by a completely new set of companies. So at this
point, the PC companies should be going into a secular decline,
which is apparent with companies like Compaq. And you have
to ask whether the same thing is going to happen to Intel and
Microsoft.
Q: Steve, let's talk about the fundamentals.
A: There are two key questions. One is: What do the
fundamentals look like today, and how do they look going
forward? And the second is: What are we paying for those
fundamentals? In the spring, some people thought we'd seen the
worst. But we thought things could and would get worse, which
they did. At this point, though, it's hard to argue that the
fundamentals will get much worse. So the focus is going to shift
to what the recovery is going to look like. And I'd argue that it
will be U- or W-shaped, rather than V-shaped.
Q: We're not going to get a big bounce, in other words.
A: No, it's not likely. Technology demand closely tracks
corporate capital spending. Over half of U.S. capital spending
goes to technology goods. It turns out, in fact, that there is very
good correlation between relative tech-stock performance and
capital-spending cycles. We're moving into a down cycle for
capital spending, and these cycles tend to last seven or eight
years. We had an eight-year upcycle, from 1992 to 2000. It's
unlikely we're going to have two difficult years and then be off
to the races. The excesses that developed over the last cycle are
going to take a number of years to work off.
It could be like the mid-1980s; tech underperformed the market
from 1984 through 1989. We're not necessarily predicting five
years of underperformance, but our intuition is that it's likely to
be longer than people think. Lower interest rates help, in that
they tend to lift corporate profits, and when profits improve, so
can capital spending. But I'd expect corporate profits to remain
under pressure, certainly through 2002. Remember, the depth of
this downturn is unprecedented: Commerce Department data
show we're down 40% year-over-year in tech orders.
Q: Yikes!
A: Also, operating margins for tech companies in the aggregate
are at a 20-year low, and capacity utilization is at about 61%,
which is historically low. There's bound to be some revision to
the mean. You should see sequential improvement as we go
through next year. But I don't expect the kind of snapback we
got accustomed to in the 'Nineties.
Q: And you'd argue that tech earnings estimates are too high?
A: First Call numbers show analysts expecting a 50% jump in
profits next year. That's come way down, but it is still somewhat
aggressive.
Q: What would be more realistic?
A: Maybe 20%-30%. You will see some fundamental
improvement next year, but it will be disappointing to many
analysts, and the time it is going to take also will disappoint. Our
latest survey of chief information officers suggests corporate IT
budgets will rise about 3% this year, and at best 2% next year.
And the numbers for 2002 have been coming down: it wouldn't
surprise me if IT spending next year was flat to down.
Q: Steve, how much did the September 11 attacks hurt the tech
sector?
A: On September 11, we were already in a global IT recession.
But in August, we were beginning to see data that suggested we
were bottoming. The OECD [Organization for Economic
Cooperation and Development] leading indicator, which leads
tech investments by nine to 12 months, was beginning to show
year-over-year improvement. The National Association of
Purchasing Management data was also showing significant
improvement. Another early warning indicator we use is
semiconductor unit shipments -- not dollars, but units. Those
peaked early in February 2000, well before the stocks peaked.
And over the summer, they improved.
Q: But the attacks derailed the recovery.
A: Since September 11, those indicators have reversed. The
OECD indicator has gotten worse, NAPM numbers have
crumbled and technology orders, which appeared to have
bottomed, are down. Semiconductor units tanked, though they
have started to recover again. September 11 messed up any
hope of a 2001 recovery.
Q: More recently, there have been some hopeful signs.
A: It's true, some companies are seeing sequential improvement
in October and November relative to September, when people
basically stopped doing business for about two weeks. So we are
getting back to "normal levels." But these levels are extremely
low.
Q: The attacks are actually boosting some businesses, right?
A: Obviously, disaster-recovery spending gets a boost. Also
security-software companies clearly benefit. We also think voice
over IP, or Internet telephony, is going to get a boost. During the
crisis, the 'Net was really the only way to communicate in
Manhattan.
Q: Voice over IP provides users cost savings, too, right?
A: There's about 30% savings over the switched network, which
is interesting, but given that the quality of service isn't as good as
wireline, it hasn't been enough to get a lot people to move. But
now resilience takes on a higher priority.
Q: Who benefits if voice over IP picks up steam?
A: Cisco Systems, main- ly. Cisco is a large company and very
much needs large new markets to go after. And voice over IP
could also be significant for them.
Q: Computer-security stocks have been on a tear
post-September 11.
A: Virtual private networks, firewalls, intrusion-detection
software; demand for all of those is going to increase. But the
thing to watch is that it takes time to roll out security throughout
a corporation. While the stocks have had a run, there are very
high expectations in terms of near-term revenue improvement.
So there may be some short-run disappointments. But if we had
a significant pullback in those stocks, we'd use it as a buying
opportunity.Internet Security Systems is a stock we like, and for
the long term, Check Point Software.
Q: Shouldn't the storage companies get a boost?
A: EMC's data-mirroring software worked brilliantly during the
crisis. And at the margin, the crisis will create more demand for
storage, but not enough to overcome EMC's current problems in
demand and pricing.
Q: Steve, there was a lot of hope that
Windows XP would drive a pickup in PC
demand in the fourth quarter. Is that
happening?
A: Not exactly. Consumer PC demand is
going to be quite weak. There's no killer
application, and consumer spending is
clearly at risk. On the business side,
we're coming up to a three-year upgrade
cycle, where a lot of companies upgraded
for Y2K in 1999. In our corporate IT
survey, we found 70% intend to upgrade
some PCs next year, and almost a third
said that they had an interest in XP. That
would suggest you could have a pretty
powerful PC upgrade cycle in 2002's second half. But I'm still
skeptical. At this point, PCs are not a strategic purchase; they're
deferrable. Most companies are lengthening their upgrade cycles.
Q: So the three-year cycle becomes four years.
A: We're forecasting 10% PC unit growth and flat revenues. In
our framework, the PC was the last wave. We don't see it as the
key driver for the industry. It is not going away; mainframes
didn't go away. In terms of what drives the industry, long-term,
it is communications. To be bullish on tech, you have to be
bullish on communications. And that's a problem now,
particularly on the service-provider side, where capital spending
is likely to be down 12% next year and down a bit again in 2003.
Q: Ugh! So no recovery in telecom spending until '04?
A: From the service-provider side, that's right. They're so
overbuilt. Enterprise spending could improve sooner, which is
why while we'd underweight communications equipment, we do
like data networking. We think ethernet is growing in importance
and that companies like Cisco, Extreme Network and Riverstone
are positioned to take business from Nortel, Lucent and Alcatel.
But we'd avoid other telecom-equipment areas.
Q: What about wireless?
A: It's an immature technology. The only successful one on the
wireless side is NTT DoCoMo in Japan. They've figured out
how to share revenues among the carriers, the application
vendors and so forth. We haven't done that yet in Europe or the
U.S. So we'd like to be positive on a company like Openwave;
they'll be well-positioned when wireless takes off. But the
market is too immature.
Q: You're not a big believer in 3G,
then?
A: We've been very negative on 3G [third-generation wireless]
for some time, and not only because of the financing issues, with
the European vendors paying so much for the licenses. It's even
farther off than people think. And when it comes, we think the
primary driver will simply be voice; 3G provides more capacity.
The applications people talked about Internet data applications
requiring 3G, which doesn't make sense. If you're doing a
high-speed data application, it's probably visual. Meaning you're
looking at a screen. And if you are looking at a screen, you are
probably not driving down the highway.
One wireless technology we are very bullish on is 802.11, or
wireless ethernet. You can be surfing the Web as fast as your
wired desktop. With Intel backing 802.11 for home networking,
it's likely to be very successful, starting this Christmas.
Q: Who benefits from 802.11?
A: It is difficult to play in a pure way, but Intersil would be the
primary beneficiary. They have a 60% share in the chip sets that
go into these access points. Cisco and Lucent sell the access
points; Texas Instruments also provides chip sets.
Q: Speaking of chips, Steve, some people have gotten excited
about a spurt in DRAM prices, on the theory that higher
memory prices could be a harbinger of a wider recovery. But
you don't agree.
A: The situation in the DRAM market is very particular and
peculiar. It's true, we have seen a jump in DRAM pricing. And it
has gotten some people excited, that is the moment we have all
been waiting for. I don't believe it. Long-term contract prices
haven't changed much. It's much more about the spot market.
We don't think there's enough end-user demand to sustain this.
Q: There's a theory that in the post-9/11 world, consumers will
nest and buy themselves a lot of electronic toys, like DVD
players and the new videogame consoles from Microsoft and
Nintendo. Do you buy that?
A: Well, I wouldn't overplay the point, but there is truth to the
fact that some people will spend more time at home. That
suggests there could be increased demand for electronic gaming
and cable. One way to play the trend is Scientific Atlanta, which
benefits from an increase in video on demand and other
sophisticated services through the cable box. And the gaming
device business is at the start of a five-year cycle, with the
introduction of Nintendo's Game Cube and Microsoft's X-Box.
But the stocks have anticipated that. Over the next few months,
software companies like Activision and Electronic Arts might sell
off and provide a buying opportunity. I'm not sure I'd chase
them now, though.
Q: Steve, Let's talk about valuations.
A: I'm beginning to wonder if we learned much from the bubble.
If tech goes up much more, we may be entering bubble 2. Look
at our tech index, the Merrill Lynch 100. About a third of its
members are losing money. The two-thirds that are making
money are selling at about 50 times next 12 months' earnings.
That's high. Now the counterargument is that tech is cyclical,
and therefore P/Es are going to be high near the bottom of the
cycle because earnings are depressed. There is some truth to
that. But, in general, I just don't buy it. In our tech index, only
about six companies out of 100 sell at under 20 times next year's
earnings. On a P/E basis, tech stocks are downright expensive.
Q: What about the coming recovery?
A: Investors are over-discounting it. Since many companies
don't even have earnings today, look at price-to-sales ratios. The
price-to-sales ratio for tech in the aggregate is about 3.5 to 4
times, depending on how you measure it. If you look at recent
trough levels, say in '96 and '98, that doesn't look so bad. But
we'd argue that conditions are more like the mid-1980s. Back
then, the top 50 tech stocks bottomed at 0.8 times sales.
Q: You don't think we are going that low, do you?
A: No, but there's enough risk to scare me. The reason we
probably don't get down that low is that there is a lot more
software today.
Q: And software tends to have higher margins then hardware.
A: Right. But with operating margins at a 10-year low, what are
current sales worth? You could argue that tech company sales
are worth less than their value of five or 10 years ago.
Q: What sectors look riskiest?
A: Semiconductors are among the most expensive companies
today, on either next-12-month earnings or trailing sales. The
semis have done surprisingly well this year, given that we
haven't yet seen the turn in the cycle. We've found that when
fund managers want tech exposure this year, they've gone to
semis. They are perceived to be early cycle plays. The fact they
have cycles is considered a good thing, because the first question
these days is whether a company will survive. And in software
and networking, you can't be sure. But in semis, companies like
Applied Materials or Intel or Analog Devices aren't going away.
They've managed through cycles before. So in tech portfolios,
there's been an overweighting of semis, which is getting
overdone.
Q: What about other sectors?
A: Storage stocks, which we're very bullish on long term, are
fairly expensive. On the other side, the computer-services stocks
have outperformed, but they're still not that expensive, selling at
about 24 times next 12 months' earnings. Computer hardware
companies actually are fairly inexpensive, but perhaps for good
reason, with the PC cycle still fairly poor. Another area that's
somewhat reasonable is the supply chain, the distributors and
contract manufacturers, companies like Solectron, Jabil Circuit
and Avnet.
Q: So what should investors do?
A: Technology is about 18% of the S&P; I would be slightly
underweighted. I don't think tech is going to lag the market to
the degree it did over the past 18 months, but I don't think its
going to outperform, either. In the short term, this rally could go
on for a few more months. But it is going to falter as you get
into next year. I'd begin selling into the rally.
Q: Where should investors go?
A: I'd start with computer services, companies like First Data,
Affiliated Computer Services and EDS -- very defensive names.
As I said, we also like the supply- chain stocks, and some of the
contract manufacturers. Most semis are too expensive, but we
do like the analog companies, like Maxim and Analog Devices.
Q: Anything else worth buying?
A: We like some semiconductor capital-equipment, companies in
the $500 million to $1 billion range in market cap. Companies
like Entegris, Emcore and MKS Instruments. The valuations on
those, while getting stretched, are better than on Applied
Materials or KLA Tencor. In software, I'd use some of the
infrastructure management software companies, like Verisign,
Mercury Interactive and Precise Software Solutions. They have
lower average selling prices on their products, which makes them
more likely to get in under the IT budgetary radar. We're also
believers in "webification," using Internet technologies to connect
applications first internally and then between companies. Our
favorite is BEA.
Q: How about something way out there?
A: If you ask: "What's the next really big thing looking out 30
years?" the answer is nanotechnology. Essentially, it's making
things 100 nanometers or less, thinner than the width of a cell
membrane. We're talking about building things up from atoms
and molecules. Within the next 10-20 years, the semiconductor
business is going to switch to atomic-layer deposition, building
up chips, molecule by molecule.
Q: Is there a way to play this now?
A: Like the Internet, nanotech is an enabling technology; it won't
be an industry all its own. But one place you'll see it first will be
in disk drives. IBM is working on a product called Millipede,
which is supposed to come out in two to three years, with 40
times the performance of current disk drives.
Veeco, the semiconductor capital-equipment company, gets
about 12 % of its revenue from atomic-force microscopes, tools
used to test and to begin to build nano-tech products. Some of it
is very far out, but there are elements you'll see in two to four
years. |
