At $16.87, Comcast is trading at 6.0x 08 EBITDA (using expected 12/07 debt of $31B). It should deliver free cash flow (fully burdened for growth capex) in 2008 of $0.65-$0.75 per share.  Assets that generate no EBITDA are worth approximately $1.50-2.00 per share.  If they grow EBITDA over the next five years at 5% annually and maintain 2.5x Debt/EBITDA, they would create an additional $0.45 per year of distributable cash, which means Comcast is trading at about a 7-8% yield on distributable cash flow.  If 5% EBITDA growth translates into 6-7% fcf growth (due to capex growing more slowly than EBITDA), implied returns for a long-term holder equal 13-14%.

 

Now let’s assess the claim that actual results in the next five years will likely be at or above those just described.  In what follows, I am not suggesting the most likely case but the most conservative one that seems plausible.

 

TV households grew at 1.6% 1998-2005 (to 110 million) driven largely by growth in occupied housing of a little over 1%.  The pay TV household market grew 3% per year (to 94 million homes), as pay TV steadily increased its share of the TV market.  Annual growth of 2-3% in the pay TV market over the next five years means the pay TV market would grow by 10-15 million homes.  Slower household formation in the next several years due to the housing downturn (relative to the 1998-2005 period) may be offset by the Feb 09 FCC requirement that all televisions be able to receive a digital signal; many non-pay TV households may sign up for pay TV rather than acquire a subsidized digital converter.  10 million new pay TV households is a conservative estimate of market growth over 5 years.

 

Assume AT&T rolls out U-Verse to 30 million households and Verizon rolls out FiOS to 18 million over 5 years.  Now assume both get 25% penetration.  This is, roughly speaking, what both companies claim they will do.  While there is some probability they will do better than they now target, the hurdles to even achieving these results are material.  Bear in mind that the scalability of U-Verse’s video technology is still not conclusively proven; they ended Q3 with 126K subs and it seems unlikely that they can produce an equivalent product to FiOS while spending, as they are, dramatically less on capex.  Nevertheless, T claimed 12/07 they were adding 10K video subs per week and would move to 40K subs per week by 12/08.  Giving T the benefit of the doubt, assuming VZ & T together achieve 12 million video subs seems reasonably conservative.

 

A conservative estimate of satellite’s likely share is harder to gauge, as constant net adds, combined with constant churn, must result in lower net adds every year.  Together DTV & DISH added 2.4 million, 3.3 million, 2.3 million, 1.9 million and 1.9 million (estimate), respectively in 2003-2007E.  Neither has shown any ability to materially increase gross adds or lower churn since 2004.  Given constant churn, each year of sub growth adds further downward pressure to net adds.  Competition from VZ & T make it unlikely that either gross adds or churn sustainably improves for either player.  If that is the case, net adds must fall to about 500K per year for each by 2009.  Assume DISH & DTV take 5 million additional subs over the next 5 years. 

 

In sum, a conservative case assumes that the pay TV market grows by 10 million, but non-cable grows by 17 million, so cable loses 7 million subs.  Comcast currently has 37% share of cable market.  Losses are likely to come disproportionately from weaker cable players, but let’s assume Comcast give up share pro rata, losing 2.6 million subs by 2012, or about 2.2% per year.

 

Video ARPU increased 5% yr/yr in Q3 07, down from increases of 8%, 6% and 6%, respectively, in Q3 06, Q3 05 and Q3 04.  These increases are driven by price increases (themselves driven by programming cost), and increased penetration of higher cost video products (digital, DVR, HDTV).  Entry by VZ & T will likely inhibit video ARPU growth.  It is too early to say how aggressively the RBOCs will price, though early indications from Verizon seem modestly encouraging.  Also, because of their small base, both T & VZ will continue to operate at a disadvantage in programming costs, which should dampen their willingness to price the product below the levels now set by cable and satellite.

 

If competition drives video ARPU increases down to 3-4% annually, then video revenue growth would eventually slow to 1-2%.  Over the long term ARPU increases below nominal GDP seem unlikely -  programming costs will continue to be passed on and continued technological innovation should permit Comcast to charge more for improved products.  Share loss should eventually abate as VZ & T achieve their targeted penetration rates and the industry reaches a new equilibrium.  As this happens, overall video growth should return to the assumed 3-4%. 

Comcast should do to the RBOCs in voice what we are assuming the RBOCs do to it in video - 25% penetration of homes passed, which for Comcast would mean roughly 50% of basic subs.  Cablevision now sells phone service to 50% of its basic video subs and continues to increase that penetration rate.  Comcast targets 23% penetration of homes passed by 2009 (11 million subs vs 3.4 million on avg for 2007).  All in it is conservative to assume Comcast’s phone revenues should at least triple by 2011 relative to 2007. 

 

HSD penetration is now 54% of Comcast’s video base.  While the U Verse and FiOS product should be as good as cable’s HSD, DSL is inferior and has begun to cede share.  Assuming Comcast eventually penetrates 75-80% of its customer base (Cablevision is now at 71%), Comcast should continue to grow subs by low double digits in the near term and low single digits beyond the next 2-3 years.  Doing so, however, will pressure ARPU, as Comcast lowers price to pick off more resistant adopters.  To date, Comcast has maintained flat HSD ARPU, but Time Warner’s HSD ARPU slipped 2-3% yr/yr in Q3 07 and Comcast’s probably eventually will as well.  A near term revenue growth in the low double digits giving way to low single digit growth further out seems conservative.  So far, wireless HSD products (Clearwire) or broadband over power lines do not pose material risks to pricing or sub count, but there is some possibility they will in the future.

 

Comcast expects this to be a $2.5 billion revenue business by 2011.  While we don’t have data to evaluate this claim, and management’s credibility on its own has been compromised, my own anecdotal observation is that the RBOC data and voice products to small businesses are significantly overpriced, about 2x the pricing for the same product to residences.  Cable should be able to move in under this pricing umbrella easily and still maintain the 50% margin they target.

 

Management claims they can add $1 billion in annual revenue by 2011 through interactive advertising to internet customers (search, interactive ads, display ads).  If so, it would probably be high margin, but they have not proven this concept yet and a conservative set of projections would not give them much credit for this.

 

Management claimed in early 07 they would grow revenue 12% per year 2007-2009.   They will do 10-11% 2007 and 8-9% Q4 07.  Add up everything I just said and you get 7% revenue growth CAGR through 2012. 

 

EBITDA margins expanded over the last several years and management predicted (March 07) that they would continue to expand 2007 - 2009.  RBOC entry may inhibit price increases and will certainly drive higher marketing expenses.  Also, HSD’s rapid growth over the last several years shifted mix toward a product with much higher than corporate-level margins.  The growth driver over the next several years, telephony, may not offer comparably high margins.  All in, it seems safer to assume flat EBITDA margin or some margin compression than expansion.  If revenue grows at 7% per year through 2012 and margins compress by 300 basis points EBITDA would compound at 5% per year.

 

Comcast’s inability or unwillingness accurately to predict its own capital needs is now justly infamous.  There is a crucial difference, however, between capital prediction and capital allocation.  Listening to management capex predictions is obviously pointless.  However, that does not mean they have misallocated capital invested in the physical plant (including digital boxes in the home).  EBITDA, net of capex in 2007 will be approximately $5 billion.  Comcast’s net physical plant (the only asset they have to replace) at the beginning of the year was $26 billion (assuming “Other intangible assets” needs replacement).  A 19% pretax return on capital for a business that can be levered 4-5x EBITDA suggests capital invested in the plant is well spent.  Further to this point, Comcast compounded EBITDA/share at 11% per year 1997-2007; had they not delevered the company over that time period, the CAGR would have been 15%.  Examining increases in EBITDA for the last 3 years relative to capital spent would support a similar conclusion – when Comcast has invested in its own plant rather than acquisitions, its results are satisfactory. 

 

Cablevision, which is approximately 2 years ahead of Comcast in penetrating its customer base with phone, HSD and digital (the last of which is the most capital intensive), saw a 16% absolute decline in its cable capex this year, to about $200 per basic sub.  This was driven largely by the fact that it reached a critical mass penetration of the capex-intensive digital product (Q4 06 was 78%).  Moreover, Cablevision’s capex decline came in the face of RBOC video rollout that already covers 23% of its footprint.  Comcast will spend about $250 per sub this year but get to a 78% digital penetration mid 2009, at which point they should start to get some relief from capital spending pressure.

Overall on capex, there is a significant chance that it will not increase materially from the 2007 level of $6 billion and may even decrease, but given the company’s history and given that RBOC entry will also push capex spending higher, it seems wiser to assume continued capex increases, albeit at a slower rate than growth in EBITDA. 

 

If capex simply grows 100-200 bps slower than the 5% CAGR we are assuming for EBITDA, free cash flow per share would grow at 6-7%.

 

If you’ve made it this far, I am, frankly, surprised.  I doubt any of the 3 people who visited my AET writeup last year made it to the end and I expect less interest in this idea. 

 

Regardless, a final comment on timing.  While a long-term shareholder should do well from here, you may well get a chance to buy this business at a better price.  The housing downturn is obviously ongoing and will continue to pound their basic sub numbers for at least several quarters, as well as pressure their HSD adds lower.  DTV’s stepped-up promotion of its high def offering will make itself felt more in Q4 than Q3.  DISH had a rotten Q3 for net adds and can be counted on to return with a vengeance in Q4.  Moreover, the RBOCs will add video subs in Q4 at an increased rate relative to Q2 or Q3.  While capex should eventually conform to the assumptions mentioned above, the CFO Angelakis on a 12/07 conf call sounded like capex might actually increase materially in 2008. 

 

More fuzzily, I do not think Comcast management has yet transitioned the company into the fighting machine it will eventually learn to be.  Despite much talk about how competitive the world is, they have enjoyed relatively benign duopoly/oligopoly status for years, in which the primary operational challenge was just to meet rapidly growing demand for utility-like essentials, not to market products and service customers in such a way as to win in a predatory competitive landscape.  Reading between the lines of Roberts on the last quarterly call, he basically admitted as much.  If you like this idea, buckle your seat belt.