2007 | 2008 | ||||||
Price: | 38.26 | EPS | |||||
Shares Out. (in M): | 0 | P/E | |||||
Market Cap (in $M): | 3,600 | P/FCF | |||||
Net Debt (in $M): | 0 | EBIT | 0 | 0 | |||
TEV (in $M): | 0 | TEV/EBIT |
Sign up for free guest access to view investment idea with a 45 days delay.
In hopes of generating a heated discussion, I’m proposing that at these prices First Marblehead (FMD) offers multi-bagger potential with arguably little downside risk. I think in a base case scenario FMD is selling for less than 6x fiscal 2010 EPS (June fiscal year). I believe that the bear theses, while very valid, are either already priced in, won’t always have relevance, or ignore other positive underlying developments.
The Bear Story
FMD is a low value-add business with unsustainably lucrative economics. FMD paid $10 million for its core asset six years ago, and now sports a $3.6 billion market cap. 52% of FMDs revenue comes from Chase and BofA. The Sallie Mae buyout means that Chase and BofA will divert volumes from FMD to SLM. Their gain on sale accounting is aggressive and they’ve already started backpedaling on their assumptions. Growth in the private consolidation loan market will permanently change prepayment experience. Federal PLUS loan budget increases will put a dent in 2008 private loan volumes.
The Bull Story
The market is a rapidly-growing duopoly. FMD is over 20 years ahead of potential new entrants. Run-rate Chase and BofA are less than 40% of revenue (and a lower proportion of EBIT), and rapidly shrinking as the higher margin base ‘non-Big 2’ business will grow over 100% in fiscal 2007. Chase/CFS (70% of Big 2) is locked in until March 2010. All three parties to the SLM buyout have said publicly that no volume will be diverted to SLM, more importantly it makes no economic sense to switch. The non-Big 2 business alone is worth substantially more than the current enterprise value. FMD has begun to originate volume in-house at substantially higher margin. Management has bought back $130 million of stock in the last two years and have earned a roughly 30% CAGR on their purchases even after the recent 40% sell-off, and they have an authorization to buy another 10% of the company.
Background
FMD is essentially an outsourced underwriter and securitizer of private student loans. FMD designs client lending programs, writes loans to those standards and packages them for sale. FMD holds no loans on their balance sheet. Bank partners are paid a marketing fee for originating the loans.
FMD owns the most extensive database of private student loan data. The database was acquired from TERI in 2001. TERI is the oldest and largest private guarantor of private student loans – their data goes back to 1985.
The private student loan market is fairly undeveloped. The market is up from $1.3 billion of originations in 1996 to $17.3 billion in 2006. Private loans can be marketed via a school’s financial aid office (school channel), or directly to borrowers (DTC). School channel loans are cheaper for borrowers because lenders must go through a bidding process in order to have access to the school channel. It also costs less to acquire borrowers in the school channel, so the savings are passed on.
Revenue Model & Accounting
Securitizations generate three revenue streams for FMD. About two-thirds of the fees are paid upfront in cash. The other third is in the form of residuals and deferred advisory fees. Both the residuals and the deferred advisory fees are the NPV of future streams of cash flow. The residuals are 80% of the non-cash portion.
Below is a standard model for an FMD securitization:
Collateral 100
Funding 133
Third Party Fees:
Marketing Fee 5 Paid to bank partners
Guarantee Fee 7.5 Paid to TERI
TERI Resid. 1.25
Total 13.75
FMD Fees:
Upfront Fee 13
Residual 5
Deferred Fee 1
Total 19
TOTAL FEES 32.75
The residuals and deferred fees are essentially excess cash flow (net interest margin) certificates that throw off cash once repayment begins. It’s inarguable that the residuals have some value (worst case is zero since the trusts are non-recourse), though exactly what value is where intelligent people can disagree. Cash flows are ideally projected as accurately as possible and discounted back (just as one values companies), so one can argue that the assumptions are aggressive, but not that gain on sale is inherently aggressive. I recommend a read of tbone841’s writeup of DFC for an excellent gain on sale discourse. The only difference between gain on sale and portfolio accounting is that with gain on sale, management has already put a multiple on cash flow, so future GAAP earnings can only be generated through additional securitizations, instead of generating a stream of cash flow over time that we’d put a multiple on. Were FMD to use portfolio accounting instead of gain on sale they’d have to make all the same assumptions except the discount rate.
Some have tried to compare the assumptions with mortgages or other lending products but the products seem completely incomparable to me for loads of obvious reasons.
Below are FMD’s assumptions versus SLMs private student loan assumptions:
FMD SLM
Default Rate 9-10% ???
Recovery 40% ???
Net Defaults 5.4-6% 4.69%
Prepayments 8% 6%
Discount Rate 11% 12.5%
So FMD’s assumptions seem more conservative than SLMs, save a slightly lower discount rate. Though one can’t know the more important variables underlying these assumptions (majors of borrowers, GPAs, etc.; FICO scores for 19 year-olds seem specious at best), so even this exercise seems only somewhat productive. At any rate, I haven’t heard a lot of accusations that SLM overstates earnings. Private equity apparently doesn’t think so, but quite honestly I’m not smart enough to argue convincingly one way or the other – if any of you can I’ll rate your ideas 10s forever. The assumptions have to be way off to kill the idea, in my opinion.
Beginning in 2007 FMD raised its constant prepayment rate assumption, confirming concerns that they were low-balling the assumptions. The fear is that growth in private consolidation loans is driving permanently higher prepayment rates, which is a legitimate concern. Management contends that a flat to inverted yield curve is driving more prepayments than are consolidations. Parents, who in over 80% of cases cosign the loans, feel that they can take out a home equity loan at 750 bps to pay off their kid’s school loans at 990 bps. I can’t add any real value to this discussion, except to say that the vast majority of FMDs loans are 20 year loans. It would be hard to get a lower payment on your typical 15 year consolidation loan. I think this is the primary concern for most borrowers, rather than cost of funds; I suspect most borrowers don’t know what their loans cost.
The sell side and shorts would also note that SLM’s prepayment experience has gotten worse. SLM and FMD management claim that a large part of this is due to in-house consolidations as SLM seeks to refinance their borrowers into longer-term products to fend off third party consolidators. They also note that many times private consolidations are a bad proposition because origination costs can range up to 700 bps, and in most cases there are no significant interest savings unlike with federal consolidation loans.
Competitive Advantages
1) FMD makes 20 year unsecured loans to borrowers with no credit history that don’t go into repayment for five years, so there are serious complexities to writing this product intelligently. Compare with auto loans or mortgages where losses often peak in 12-36 months.
2) FMD’s 20 year head start allows them to price loans more efficiently, pay lower funding costs and pay lower guarantee fees than any potential entrant could.
3) New entrants would need to earn the confidence of the rating agencies, guarantors and ABS investors. It’s worth noting that neither Chase nor BofA has ever executed a securitization on their own.
4) FMD is a marketing platform with a network of both borrowers and investors that is rivaled only by SLM. The platform seems somewhat reflexive.
5) It requires substantial scale to package a securitization of worthwhile size. Granted, this is somewhat less relevant for the Big 2 than the smaller clients.
6) Banks don’t want to hold illiquid, non-investment grade residuals that don’t cash flow for five years, and it’s tricky to sell securities BBB and deeper to cash them out – FMD only recently began to on a small scale.
I think the fees are so high that it’s a turnoff for a lot of people looking at FMD. I think it’s inarguable that fees will come down eventually. Marketing fees to banks could increase. Or pricing could come down and not be offset completely by lower funding costs as the market matures. At that point the product would obviously become more accessible to some borrowers, fueling further market growth. Outside of egregious legislation, I can’t imagine a scenario where fees would fall suddenly and precipitously, but I’m open to suggestion. As long as I’m right, revenue can still grow very healthily since origination growth would outpace percentage fee compression.
I would also note that fees have increased in recent years. All-in fees:
6/07 3/07 12/06 9/06 6/06 3/06 12/05
Margin 18% 19% 20.1% 17.8% 15.3% 15.5% 14.9%
DTC 80% 74% 93% 70% 82% 70% 73%
Note that the 12/06 securitization benefited from a 93% mix of DTC, which are substantially higher margin than school channel. The 6/07 deal is the first inexcuseable crack in margins. The reason for the lower 6/07 margin is because it was the first deal that the rating agencies used the new 8% prepayment rate assumption. Higher prepayments means there’s less projected cash flow available for debt service, which decreases the amount of debt issueable against the same dollar of collateral. Less debt issuance means FMD takes out less cash upfront, and gets more residual value. Because the residuals are discounted at 11% and the notes are effectively discounted at 5.6% it’s not a one-for-one trade-off on day one, so all-in fees look lower.
Otherwise, to anticipate a question, I think the 2007-2 deal was good. Funding costs were up 2 bps, but spread was up 2 bps. Astrive volume was probably exceptional given the contribution from Charter (Charter is the bank that FMD runs Astrive volume through, until FMD acquired a bank). Reserve account requirements were down substantially – from 23% of trust assets to 19%. The reserve account is an excess bond issue that goes toward a sort of sinking fund so that the trust can cash flow while the loans are in deferral. Lower reserve account means lower interest expense and more residual value. The reserve account is somewhat subjective because it has a lot of conservatism built into it. I think ordinarily they’re sized to hold up under 18% LIBOR.
The 2007-2 deal was worse sequentially, but to put it into perspective here are some stats from past deals:
6/07 3/07 12/06 9/06 6/06 3/06 12/05 6/05 3/05 12/04 6/04
Parity 96.6 95.6 96.2 95.2 98.4 98.4 99 99.4 98.7 98.9 108.7
Loan/Liab. 74 69.4 70.7 75.2 75.6 83.1 76.6 76.8 75.2 72.9 87.8
Reserve Ratio 18 22 21 15.3 19 11.1 18.9 18.8 18.8 21.8 16
As long as you believe management’s yield curve story, there’s no reason for CPR assumptions to change incrementally (except fraud of course) as the yield curve seems to be steepening. CPR is the most important assumption. Margins are more levered to discount rate, but discount rates are less volatile than CPR. See page 28 of the latest 10-Q for a sensitivity analysis.
In large part I think the long-term trend of percentage fee growth is a byproduct of a growing mix of non-Big 2 revenue, which carries a lower marketing fee, and thus higher margin. I suspect fees paid to Big 2 bank partners have actually increased in this time.
I suspect the new Astrive product also plays a material part in the increasing fees. Astrive is the same product only it’s originated in-house, with no bank marketing partner. One can deduce an extra 5% on a 19% spread can do wonders for margins. Astrive has grown from zero 18 months ago to 10% of revenues today. It’s my suspicion that looking out three years Astrive could become a large contributor to volumes, eliminating third party dependence and boosting margins considerably – management won’t advertise this obviously. Non-Big 2 revenue will also continue to drown out the Big 2 business. Big 2 concentration over recent history:
2007E 2006 2005
44% 52% 65%
SLM Takeover
As for whether Big 2 volume will be channeled to SLM post-takeover, I think it’s likely that it won’t.
1) Chase and BofA will each own 25% of SLM. If instead of the guaranteed 5% upfront cash filet, they decided to go for the whole fish, they’d each take 25% of 24% (FMD’s fee plus the marketing fee), or 6%. They’d end up with slightly better economics, but it’d be 4.5% cash and 1.5% deferred.
2) It looks bad from an antitrust standpoint for a deal that is already being rigorously scrutinized by regulators. Combined the three parties control roughly two-thirds of the private student loan market.
3) For every extra dollar that either party makes, they’re supplying their biggest competitor with the same dollar.
4) The Chase contract is in place until March 2010. More importantly, the contract stipulates that if Chase wants to originate volume outside of FMD it has to be originated under another brand, prohibitively burdensome in my view.
Model & Valuation
Note: My numbers exclude Processing Fees from TERI and Admin & Other Fees since the TERI fees are zero margin, and Admin & Other are immaterial. Revenue, operating expense, and EBIT % won’t tie, but EBIT $ will. I wanted to make it easier to model for this write-up.
I’d like to preface this section with my overriding thesis on this stock. I think of FMD as two businesses – Big 2 and the base non-Big 2 – to help eliminate noise about the Big 2 risk. I believe the base business is a very valuable franchise. The base customers are smaller banks with no ability or willingness to structure securitizations on their own, and no bargaining power versus FMD, with all the competitive strengths detailed above. This base business is growing parabolically. I think if the two businesses were separately-traded the base business would easily garner a 30x multiple.
I think the Big 2 will generate about $360mm of revenue in 2007, putting the base business at about $320mm of revenue in 2007, and maybe $500mm in 2008 (my estimate). Last time FMD was doing around $320mm of revenue they had an 80% EBIT margin, and I believe there’s no reason that as a stand-alone business the base business couldn’t do the same, probably better because of the higher margins in the base business and Astrive. Obviously it would take time to either scale the overhead down, or grow into current capacity.
Back-of-the-envelope:
2007E 2008E Worst Case (Using ’07)
Base Rev 320 500 320
EBIT % 80% 85% 70%
EBIT 250 425 220
After-Tax 150 250 130
Multiple 30 25 15
EV 4,500 6,250 2,000
Shares 95 95 95
Stock Price 47 65 20
This valuation ignores $800mm ($8/share) worth of cash and fully-taxed residuals, and nearly $100mm of latent equity in warehoused loans on the balance sheet today, not to mention no value for the Big 2 business even if it only has option value. So customer concentration risk seems irrelevant at these prices. At any rate, by the time Chase comes up for renewal in 2010 the base business will be significantly bigger than $320mm. Try getting downside without ignoring Chase.
(millions) 2006 2007E 2008E 2009E 2010E 2010 Best Case
Base Volume1 .6 1.4 2.3 3.1 3.5 4.5
GOS Margin 22% 22% 21% 20% 19% 20%
Base Rev. 140 310 480 620 670 900
Big 2 Volume 1.7 2.3 2.3 2.3 2.3 2.3
GOS Margin 16% 16% 15% 14% 13% 13%
Big 2 Rev. 278 370 350 320 300 300
Total Rev. 419 680 830 940 970 1,200
Total GOS % 15.2% 18.4% 18.0% 17.4% 16.7% 17.6%
Adj. Op. Exp2 73 100 110 120 125 130
% of Rev. 17% 14.1% 13.3% 12.9% 12.7% 10.6%
EBIT 345 580 720 820 845 1,100
EBIT % 83% 86% 87% 87% 87% 89%
After-Tax 200 350 430 490 500 660
Free Cash Flow3 215 267 300 315 420
Cash, EOP 330 600 900 1,200 1,400
Resid., EOP, Taxed4 470 630 800 1,000 1,100
1All securitization volume and GOS margin assumptions are purely my guess, hence the rounding errors in 2006 and 2007. FMD doesn’t disclose these metrics.
2Adjusted operating expense is total operating expense less Processing Fees from TERI and less Admin & Other Fee revenue. This assumes the corresponding revenue is zero margin.
3 This assumes that 1/3 of revenue is non-cash. All operating expenses are cash. This calculation ignores cash flow from maturing residuals, the first of which will begin cash flowing in the next year or two.
4Residual build is 1/3 of revenue. I fully tax the book value of the residuals at 40%. I also don’t accrete the residuals for time value at the 11% discount rate.
Base Best
2010 AT 500 660
Multiple 14 16
EV 7,000 10,500
Cash 1,200 1,400
Residuals 1,000 1,100
Equity 9,200 13,000
Shares 95 95
Stock Price 97 135
This valuation ignores potential value creation from buybacks, which are a certainty.
I see no real reason for a sell-off in the short- to medium-term. BofA comes up for renewal 5/31/08, but they’re only 10% of revenue, and they’re not going to cancel before the fall and spring semesters because of the complications of replacing FMD during busy season. SLMs absence from the securitization market until the buyout closes will be good for pricing probably for the next several months. Chase is locked up until 2010. In the meantime the base business is growing phenomenally, so time is very much on our side.
show sort by |
Are you sure you want to close this position First Marblehead?
By closing position, I’m notifying VIC Members that at today’s market price, I no longer am recommending this position.
Are you sure you want to Flag this idea First Marblehead for removal?
Flagging an idea indicates that the idea does not meet the standards of the club and you believe it should be removed from the site. Once a threshold has been reached the idea will be removed.
You currently do not have message posting privilages, there are 1 way you can get the privilage.
Apply for or reactivate your full membership
You can apply for full membership by submitting an investment idea of your own. Or if you are in reactivation status, you need to reactivate your full membership.
What is wrong with message, "".