Brattle prattle…

Posted in Financials, ICO/DBSD, LightSquared, Operators, Regulatory, Spectrum, TerreStar at 11:00 am by timfarrar

Last week’s Brattle Group report for LightSquared, not only highlights the regulatory “gift” from the FCC involved in their January 2011 waiver, but also contains an estimated value of $12B for LightSquared’s L-band spectrum. This valuation is derived by assuming that LightSquared’s terrestrial-only waiver means its spectrum should have an equivalent value to “unencumbered AWS wireless broadband spectrum”, which Brattle believes to be worth approximately $1.00 per MHzPOP.

Unfortunately, this valuation, which is derived from an April 2011 Brattle Group analysis, is also deeply flawed. As an aside, it is almost identical to the valuation put on LightSquared (then MSV) by Brattle back in October 2005, when they estimated that 30MHz of MSS-ATC spectrum had a “potential” value of $0.99 per MHzPOP (although ironically, at that point it was asserted that “there are questions regarding interference and other technical issues that potentially make L-band use for ATC problematic, which would require L-band spectrum to be priced at a discount to S-band spectrum”).

The most obvious problem with Brattle’s valuation of LightSquared’s spectrum is that is is just as inappropriate to use AWS spectrum as the sole benchmark for L-band valuation today, as it was to use PCS as the benchmark in October 2005. Today the AWS-1 block has substantial existing infrastructure already deployed (both base stations and handsets) which can readily make use of the spectrum, whereas LightSquared was not even able to provide test handsets or full power base stations for the recent Las Vegas testing.

In addition, there are also more subtle issues which render the AWS value in the April 2011 paper unreliable. Notably, Brattle estimates trends in spectrum pricing by reference to a spectrum price index created and maintained by Spectrum Bridge. Brattle suggests that the index “tracks changes in spectrum value reasonably well” because as one example, “the change in SpecEx Index values closely tracked the change in AWS spectrum value based on NextWave’s AWS spectrum sale to T-Mobile in July 2008. The NextWave sale reflected a 91% increase in AWS spectrum value, whereas, the SpecEx Index in the same period indicated an 86% increase in spectrum value.”

However, the cited transaction reflects the sale of spectrum which by July 2008 could be readily put into use in T-Mobile’s built-out AWS-1 network, compared to spectrum which in summer 2006 had not even been cleared of interference. It is as if I bought some farmland and then a developer put in utilities and roads on the adjacent tract of land. The value of my particular plot might have increased, but that would say nothing about the market price of farmland. In spectrum terms, no-one (including Brattle) would suggest that Aloha’s windfall on sale of its 700MHz spectrum to AT&T was largely due to a general increase in the value of spectrum, as opposed to the DTV transition and the 700MHz auction creating certainty about whether the spectrum could be put to use.

A second factor is that of survivorship bias: in the cited NextWave transaction, NextWave’s AWS holdings were sold because the offered price was acceptable (higher than the original price paid), but NextWave failed to sell its 2.3GHz and 2.5GHz spectrum holdings which were on offer at the same time (because the offers, assuming there were any, were too low). Spectrum Bridge claims that its index takes account not only transaction data but also a “custom weighting of value, spectrum, and macro-economic based factors driven by SBI’s valuation and trading data”. However, this inevitably obscures the methodology and makes it all but impossible to determine whether the index accurately tracks spectrum values.

Returning to my farmland analogy, it would seem that the most important factors in attempting to exclude both “improvement” and “survivorship” bias would be that (for currently unused spectrum) the ecosystem for use of the spectrum should not have changed dramatically in the intervening years, and the timing of the sale should be dictated by external events (e.g. a bankruptcy auction) rather than by whether or not the prior holder can make a profit over what it paid previously. In that context, the recent sale of DBSD and the current auction of TerreStar almost certainly provide a better indicator for trends in the value of LightSquared’s spectrum than the index used by Brattle.

At current levels (of roughly $0.25 per MHzPOP), DBSD and TerreStar’s spectrum has actually gone down in value compared to 2005 and is broadly similar to the trading price (and the Motient/SkyTerra exchange valuation) back in 2006 (before most of their satellite construction expenses were incurred and well before the spectrum could be brought into use). This compares to a SpecEx index which has doubled since mid 2006 and trebled since 2005.

If (perhaps optimistically) we assumed that LightSquared’s spectrum with the terrestrial-only waiver is similar to AWS-1 spectrum when it was auctioned in 2006, then based on the DBSD/TerreStar trend (of minimal change in price since 2006) the AWS-1 auction pricing ($0.54 per MHzPOP) might be an appropriate valuation to use. On the other hand, if DBSD and TerreStar also include in their current valuation some allowance for the possibility that the FCC might also grant them a waiver (so are more directly analogous to LightSquared), then their $0.25 per MHzPOP valuation might be more appropriate.

In order to come up with an actual dollar valuation of LightSquared’s spectrum assets, you then need to take into account the impact of interference (i.e. whether to use 20MHz or 40MHz as the basis of valuation) and the NPV of the Inmarsat payments (where a relatively low (8%?) discount rate would probably be used by any established wireless operator purchasing this spectrum).

At the low end (20MHz @$0.25 per MHzPOP), you come out with a negative valuation after the Inmarsat payments (minimal value for the first lien debtholders), whereas at the high end (40MHz @$0.54 per MHzPOP) you have a valuation of about $4B (before paying off the first lien debt). That’s why I said that “It is very hard to see how you justify an equity value for this business (in line with) what Harbinger has invested, unless it can show it will be able to use all the spectrum it owns”.

1 Comment »

  1. krbarker said,

    June 27, 2011 at 4:33 pm

    The latest CMRS Competitive Report from the FCC (http://wireless.fcc.gov/index.htm?job=cmrs_reports) seems to confirm your numbers. They indicate LS will have 260 POPs by 2015, and they also used the AWS-1 auction numbers. By my math (with 40 MHz) of spectrum that’s around $5.6B worth of spectrum. However, as we have both discussed, this is highly subject to interference. What I don’t understand is how can LS keep dropping their power and amount of spectrum, and not change their valuation. In fact, I’m not sure I understand the whole base station power issue. Dropping power of a base station simply means I need more to cover the same area at the same user QoS. More base stations means more cumulative power in the spectrum creating the same issue. We’re still updating our interference analysis (www.questinygroup.com/qgi-blog), and maybe that will resolve the issue in my mind.

Leave a Comment

You must be logged in to post a comment.