GS: EMS group: good high-beta way to
participate in a likely seasonal rally
52-Week Range US$15-7
YTD Price Change -41.75%
Market Cap US$3.8bn
EPS Growth Estimate 15%
Fiscal Year (ending in Sep)
2003 2004E 2005E
US$0.05 US$0.26 US$0.50
— 28.2X 14.7X
(1) SEASONAL RALLY SEEMS LIKELY: Nobody can be sure if we're at the beginning
of the end of the cycle, or not; below we discuss why we lean a bit towards the glass being
half full. Either way, however, we think 4Q seasonality is likely to drive a tech rally by
year end (perhaps modest Sep rally, Oct lull, then strong post-election). (2) EMS IS HIGH
BETA W/ APPEALING RISK/REWARD: EMS is down 22% since 6/1, absolute
valuation is modest (16x 05E & our ROIC model shows 30% upside medium term); we
see downside risk of 15-20%. (3) DIFFERENTIATION WITHIN EMS HAS
NARROWED: EMS names have different end mkt exposures, but we think differences in
valuation & upside have narrowed considerably; our ratings changes are sector relative,
but again, we don't see significant risk/reward differences here. Our favorite names are
SANM (up to OP from IL), SLR (OP), CLS (lower to IL from OP) & FLEX (IL); small
cap PLXS (up to IL from U). SANM is our top pick to year end.
(1) SEASONAL RALLY SEEMS LIKELY
We don't have much to specifically point to other than history of seasonal tech rallies
from fall to year end, each of the last several years. Perhaps this can be explained by positive
seasonal data points being too difficult to distinguish from lasting end demand improvement. In
other words, seasonality sometimes provides a head fake, and sometimes doesn't. Either way,
stocks should rally. Separately, this analyst thinks election jitters and terrorism concerns are likely
to muddy a seasonal tech rally; our best guess is a moderate pickup in September, followed by an
October lull or pullback, followed by a healthy rally in November (if a major terrorist strike does
not materialize and a Democrat sweep of The White House & Congress does not materialize - the
latter seems highly improbable).
We lean toward a macro tech view that the glass is half full:
a) Tech's seasonal pattern seems more back ended now vs the 1990s
As we step back to reflect on merely OK enterprise h/w results for Jun-Q, we keep coming back to
a hypothesis that seasonality may be different now, more back-ended, than the 90s. The 90s had
many "must have" rollouts & upgrades (networks, new chips & O/S, ERP, email, Internet, CRM, &
Y2K); today, besides security (and some storage), most spend is "nice to have" & can wait until
4Q or late qtr negotiation yields low price. Rather than concerns we're at a peak, we're thinking that
04, like 03, may just be very back ended. Our Tech Strategy Team's Survey work (July 29, 2004)
shows some evidence to support this hypothesis, as does our additional anecdotal data gathering
from IT managers. The Survey found IT budgets to be more fluid than in the past, with 25%
holding off purchases in 2Q (nearly half of these at the request of senior management). Just over
half of respondents expect to spend their entire budget by year end. FC ests for companies in the
GSTI forecast 7% 4Q growth q-q, which is less than the 8% seen the last four years and survey
data suggesting this is achievable.
b) In hindsight, 4Q03 through 2Q04 pattern fits a mild mid-cycle correction
-- 4Q03 was so strong, component lead times lengthened & some were in shortage; OEMs had
unfilled orders.
-- Early 1Q04, OEMs directed mfg (internal, EMS & ODM) to boost inventory so OEMs wouldn't
miss sales; indeed, 1Q inventory was up at some OEMs & particularly EMS. 1Q semis responded
to higher orders & revved up internal or foundry capacity/orders.
-- 1Q end demand was good because of 4Q carryover, but it decelerated into 2Q.
-- 2Q end demand was just OK, but sufficient to burn excess mfg inventory (EMS inv. down 5%
excl FLEX); no major OEM or EMS names really came up short except those w/ high expectations.
-- Mfg inventory adj in 2Q coincided with higher semi production (see #2), so now several semis
ended 2Q with excess inventory; semi distributors see the correction now over & balanced. Based
on foundry utilization, semi prices could actually weaken, but that may be a + for hardware OEMs.
c) Semi inventories have raised the concern that the semi-cycle may be approaching a top. What we
don't know is if this is a top followed by a shallow correction or a top followed by a deep bottom.
History has seen both many times. We remind investors that historically, semi-cycles have often
been coincident with broad tech spending cycles, but not always (as was the case with the
semi-cycle tops seen in mid-95 and 97).
(2) EMS HAS APPEALING RISK/REWARD
We think EMS has appealing risk/reward profile at current levels and should benefit from a likely
seasonal tech rally (group beta is over 2). We see 30% upside medium term with 15-20%
downside. By no means do we think EMS is near trough valuation; investors who think recent data
signals the beginning of the end of this cycle should stay clear of EMS. Our valuation view is
based on current 05 ests that we think have room to move up for most names if the cycle continued
into 05; of course, we see some downside risk to ests if we have topped out now in this cycle.
From late 2003 through May, EMS names have traded from 18x-24x 05 ests, generally in the low
20s. We recognize that this included the prospect for upward est revisions, but they only came from
a few names. Current P/E on our 05E is 16x for names under our coverage; that leaves about 15%
upside if multiples expand by year end to just 18x (and that would be 18x 05 vs year ago levels of
20x 05).
Our ROIC valuation model has proven quite reliable for long term valuation purposes. On average,
our model shows 30% upside for names we cover, from now to mid-05. SANM and SLR are at the
biggest discounts to our ROIC model, showing well above average upside; CLS and FLEX show
appealing upside inline with the 30% average.
Finally, other metrics, are rather modest at current levels. Price-to- tangible book bottomed in late
02 at 1.0x; we doubt the next down cycle would get that bad. Current median price-to-tangible
book is 2.2x, with price-to-book of 1.6x. SLR, CLS, PLXS, and BHE are at or below avg,
providing downside protection; FLEX is the high at 4.1x. Most names are relatively inline with
price-to-LTM revs of 0.5x, although SANM is 0.33x and JBL 0.75x. Again, in historical
pre-bubble context, these group levels are rather modest to below historical norms).
(3) EMS STOCK DIFFERENTIATION HAS NARROWED
As noted above, we believe that risk/reward differences for the EMS names we cover have
narrowed considerably. Due to our sector-relative rating system, we must identify Outperformers
vs. In Line performers, but valuation differences are no longer as significant as our ratings imply.
Our favorite names are SanminaSCI, Solectron, Celestica, & Flextronics for larger names, and
small cap Plexus.
To us, appeal of one name vs. another is now limited to investor preferences for different end
market exposures (though even here, many exposure levels are converging). Two names with more
distinctive end-mkt exposures are Flextronics (handset and consumer tech exposure are each about
a third of revs) and SanminaSCI (~35% of revs from commercial PCs & Intel-based servers, and
15+% exposure to high-end enterprise computing).
Our ratings adjustment of CLS (lower to IL from OP) and SANM (up to OP from IL) reflects
substantial out performance of CLS shares and recent underperformance of SANM. We still see
CLS as attractive here, with likely upside est revisions. SANM, however, is the only name at this
time with a discount to peers that seems unjustified, hence our upgrade. In the case of PLXS,
following severe underperformance, its premium valuation has eroded; we think shares are as
attractive as larger peers, hence our adjustment to IL from U.
SANMINA-SCI: OUR FAVORITE NAME TO YEAR END. As noted above, upside differences
among the group have narrowed considerably, but when pushed for our favorite name, we think
SANM, for 3 reasons: (1) VALUATION - EMS group is trading at 15.7x avg P/E on CY05E with
SANM at 13.2x, and SANM shows 50+% upside over 12 months using our proprietary ROIC
valuation model vs. 30% group avg. (2) ENTERPRISE SEASONALITY: more than 50% of revs
comes from enterprise computing hardware (commercial PCs, Intel-based servers, high- end
servers, & storage); these are among the more reliable 4Q seasonally strong segments. (3)
RESTRUCTURING: A new round of cost cuts announced with Jun-Q results affords some
flexibility and cushion to EPS results & guidance.
I, Stephen Savas, hereby certify that all of the views ... |
