Description

Summary

Apollo Education (APOL) trades at severely depressed levels, with a recent market cap less than cash. A case for the stock to triple within two years can be made with reasonable assumptions, while adjusted cash and a performing international business provide robust downside protection. At current levels risk/reward seems highly skewed in favor of the long side.  

Apollo is a for-profit education company, running the University of Phoenix (UoP, 84% of TTM revenue) as well as a portfolio of other educational platforms. The UoP has been in a period of rapid and business changing decline, while the international businesses have been growing at a rate in the high teens.  

Why now?

Steve Eisman and others made a compelling (and correct) short call on the industry in 2010. The shorts emphasized the generally poor quality of education, debt burdens that were dramatically too high, and the inevitability of changes to the public subsidies that fed the machine. An investigation headed by former Sen Tom Harkin (D-Iowa) concluded the for-profit industry often provided "debt but no degree". For-profit institutions have subsequently seen casualties, including the bankruptcies of Corinthian Colleges (COCO, peak market cap of $1.5B) and Anthem Education as well as the restructuring of Education Management Corp (EDMC, $3.5B at peak) where bondholders took >95% of the company. Other for-profit educators are currently on the ropes. APOL has seen its stock plummet >90% in the past six years.   

Plenty of convincingly articulated long arguments have been made for APOL over the past few years, including a well-argued piece on SumZero in June 2013 arguing for better than a double from $19. With the stock today near $7, here's a quick post mortem: about $7 of unrestricted net cash has dwindled to $3, a perceived optionality in their proprietary learning system is gone as that tech is worthless, and the enrollment trough (author had seemingly punitive enrollment cuts to 183K/285K students in bear/base case) missed both the scale and speed of the decline.  

Thus a long thesis from here needs to answer why this catch a falling knife story won't end in bloodshed. Viewed from a different angle, this also explains why the opportunity exists: many of those who would be logical buyers of this stock probably already bought and got burned, and badly. Who's coming back for a second helping? (Apologies for the mixed kitchen metaphors.)  

Quick Sum of the Parts  

Cash ($3): Aug '14 unrestricted balance net of all debt of $458, less $109M for October acquisition of Career Partner GmbH = $3.20 per share.    

Global Business ($4-10): 2015 revenue of $391M, which management claims will grow at 20% to end of decade (organic constant dollar revenue grew at 20% 2015.) Valuing conservatively at 1x forward revenue (see below discussion), and assuming their recent Germany purchase is worth at least half what they paid, this segment is worth $4.60 per share. In an upside scenario (20% growth to end of decade, as management has laid out), Global's value is north of $10 per share.  

US Business ($3-16): management has guided to 2016 enrollment at University of Phoenix of 150,000 (which would be the largest percentage drop in their history) and bottoming out at 140-150,000. Using a lower trough enrollment at 100,000 (50% revenue cut from here) and using 1.4x for the degree of operating leverage (matches the historicals, though over time they could likely do better since their cuts have not been able to keep up with the pace of decline so far), you end up with downside EBIT around $40-50M. What's that worth? At 7x EBIT, that's ~$300, or close to $3 per share. In a less draconian case, 150K enrollments and 15% margins (at the low end of management case), 8x EBIT gets you $16.50 per share.  

Potential liabilities (loss of $0-$1): past examples of settlements for misleading students: DoE fined Corinthian $30M, FTC fined Ashworth $11M. More than tripling the like for like fines leaves you with about $1 per share downside.  

Adding up the parts you get a fairly wide range of $10-29. Though I could argue for more upside from there it's not necessary for an investment--the upside will take care of itself. Rather, the time and space seems better used stress testing the downside. Specifically, under what scenarios could the US business be worth <$0? And could fines eliminate all (or more than all) of the cash?  

US Business  

The For-Profit Value Proposition  

Right off the bat I'll acknowledge it's dangerous to approach downside analysis with a philosophical discussion, but I'm starting there anyway. Question: Does for-profit education have an ongoing role to play in the US? My answer: clearly yes, within defined boundaries. I've spent the majority of my adult life in some way actively involved with a university: as student, admin staff, chaplain (long story, for another time) and adjunct faculty. Our private universities are terrific at some things: academic research, educating certain kinds of students, protecting academic freedom, defending their value proposition, and holding lots of long meetings. They are less good or even terrible at some other things: making quick decisions, implementing change, growing rapidly, and educating certain kinds of nontraditional students.  

The core thesis: for older, working adults less prepared for a traditional postsecondary academic experience, for-profit institutions have a lasting structural advantage. Advantages are derived from a lack of constraints, which for traditional universities include faculty independence (making best practices and curriculum standardization difficult to implement), hard to change traditions such as the agrarian academic calendar, complicated decision making processes, and alumni pressures. For profits thus allow for much closer collaboration with employers, greater standardization of curricula and best practices, cheaper delivery mechanisms, and quicker response to data driven market research. This advantage has been obscured by a flood of federal dollars and highly skewed incentives, which led to very real and sometimes terrible abuses as well as creating an asset bubble. Those abuses are in the process of being corrected, leading to both massive capital destruction and significant value   dislocations. As these reforms enter the late innings, the performing core will show again that it can add value, at some fraction of the distorted peak levels.  

The attraction of a college degree is clear: a Bachelor's degree pays (see below graphic). But note the earnings disparity between 4 year and 2 year degrees. Run the calculations with whatever assumptions you want, but 30 years of the below earnings differential taxed at an incremental 30% rate and discounted at 10% has a present value of $115,000. The Associate's Degree under the same assumptions has a present value under $15,000. This is just an argument for the marketability of a 4 year degree, not an airtight argument for Apollo--you'd need to factor in graduation rates, the fact that you put in a lot of study hours you could be getting paid for, the relative value of a UoP degree relative to a "median" degree, etc.    

For those who missed the traditional age 18-22 college window, perhaps because they haven't thrived in a traditional academic setting, what are the options? 

• Community colleges can be a good option, but don't offer Bachelor's degrees, and with state and municipal budgets where they are it's difficult to see where they have any substantial incremental capacity.  

• Traditional universities haven't done well with these students. Night and weekend classes are hard to come by. Some are ramping up online programs, but are limited in scale by faculty/alumni pressures (don't dilute the value of my degree!) and budgets. o   Selective admissions schools will grow, but this isn't really APOL's target audience. One of the best and most aggressive of these is Penn State, which has committed a total of $20M over 5 years to triple enrollment over a decade (to 45K students.) There will be others, but it's going to take quite a while even in aggregate to scale up. Also, note that the Penn State World Campus acceptance rate is in the low 20%s, and their costs are slightly higher than APOL (nothing wrong with that, they have a good brand, but this keeps any potential competitive threat in perspective.)

• Competency based programs, massive open online courses (MOOCs), etc. This is an undeniable and interesting trend, but think for a minute about who they are designed for. You can get a great engineering degree at Georgia Tech for a fraction of the cost if you go online/competency based. But you are getting very little help along the way (by design) and the graduation rates are very low (single digit for GA Tech's program). APOL's model is geared toward students who need/want a little extra handholding. These alternatives aren't going to poach a lot of their students.

• Other nonprofits. It's competitive, but as we've seen some of these players are going to get squeezed out. Over time, scale gives you some slight advantage in terms of marketing, course design (basically a fixed cost per course and a little scale if you roll it out more widely), data mining (knowing patterns of when students are struggling is very useful). There's clearly no real barrier to entry here, and success requires efficiency. Some institutions (e.g. Strayer Education) are very close to being there. Apollo is on its way, with a number of the key building blocks in place.

Skewed Incentives  

The for-profit asset bubble was caused by several factors. A flood of Federal dollars in the form of Title IV funding essentially gave free money to schools for enrollment of any kind. Predictably, this skewed the incentives, with various institutions paying recruiters solely on quantity of new enrollments. The abuses in the industry were massive, including attempting mass signups of retirees in housing projects (hard to pay off debt if you don't have any working years remaining); hiring strippers to lure new recruits to enroll; the list keeps going. Apollo was not the worst offender, but they were certainly enthusiastic participants.  

One measure of Apollo's participation in the get-paid-for-no-real-value-add circus is the stunning rise of their Associate's Degree students. From a very small base (<2% of enrollments at FYE 2004), they quickly grew 2 year programs to a peak 45% of their enrollment in late 2009 and 2010. Intuitively, this expansion was not value creating for students. But just how bad was it?   One useful measure of value to students emerged from the lengthy debates about Gainful Employment (GE). Fought tooth and nail by the for-profit industry (who were benefitting from no-value-add students), these regulations proposed to measure value creation in two ways. First, for program graduates, a measure of debt service to earnings, with concerns beginning when that ratio was over 8%, and programs failing at 12%. Second, to capture all students and not just graduates, a cohort default rate would measure defaults, with concern here when this ratio exceeded 20% and failure at 30%. The final Gainful Employment regulation, effective July 2015, incorporates variations on the first but not the second metric.    

Regardless of implementation, the originally proposed regulations provide a useful proxy for value creation. Here is a list of publicly traded institutions based on 2012 GE data, listing their programs at risk along with the percentage of students in those programs:      

(Source is "2012 GE Informational Rates", raw data available at DoE website. First columns list failure/concern rates according to original GE proposal, which is a useful proxy for actual value created. The final two columns list failure/concern on final GE rules.)  

For APOL, GE regulations were a significant threat, though their sting was reduced with the dropping of the cohort default calculation. Still, there was a clear indication that the overall student value proposition was suspect.  

However, the GE data also provides useful insights into the quality differences within a particular institution, since it lists everything by individual degree program. Apollo has 144 separately listed programs, ranging from an Associate Degree for "Teacher Assistant/Aide" (3,900 students, 15% debt to earnings and 40% cohort default) to a similarly sized Bachelor in "Registered Nursing" (3,050 students, 2.5% debt to earnings and 9% cohort default). Crunching the data by credential, and measuring against the initially proposed (and I think quite reasonable) GE regulations: the Associates Degree programs in aggregate show that 95% of their students are enrolled in programs classified as fail/concern , while 9.3% of Bachelor Degrees students are in those fail/concern programs.  

This gives an idea of both the focal point of the problem, as well as its scale. Associate Degree programs in general create far less value. Bachelor Programs are not all created equal, and some need a radical change or are perhaps unfixable. For example the 2012 GE data show under the new regulations, 4 Bachelor programs totaling 5,725 students would fail. Ordered by largest number of students, this includes programs in General Human Services, Behavioral Sciences, Elementary Education, and Health Care Administration. However, the scale of the program is an order of magnitude different than Associate programs.  

Illustrative Student Economics  

Tuition is ~$60,000 for a BA degree (varies by program & transfer credits). This is offset by up to $26,000 in Pell Grants for a net cost to the student of $34,000. Subsidized Stafford loans (at 4.3%, payments starting at end of program) can cover $23,000, and let's say this student pays half the rest out   of pocket ($5,500 over 5 years) and gets a private loan for the other half. So student graduates with total debt of $28,500, and to make the math easy assume a 5% blended interest rate.  

Debt/Earnings ratios are calculated with 15 year terms, so the debt service calculation here is $230 per month. The warning zone of 8% D/E corresponds to is a salary of $34,300. If D/E is 5%, then the relevant salary is $55,200. Some of the activists have argued that the 8% D/E in the final GE regulations is too high, and I can see something to their point. But at least at the 5% D/E level, it doesn't seem like there is a lot to argue with regarding the value and affordability.  

Cancer Surgery  

Applying Charlie Munger's analogy of cancer surgery, if you cut out the bad parts is there healthy tissue left?  

From UoP's enrollment peak in May 2009 until Aug 2014 (when the company stopped breaking out enrollments by degree), Bachelor enrollments were down 10% and Associates enrollments were down 72%. Advanced degrees were down 50% as well, off a much smaller base.  

The company is guiding to 150,000 as stabilized enrollment, and no one appears to believe them. This represents a decline of 83,500 students from Aug of 2015, and we're just about exactly half way there (191K as of August 2015 report.) They had 52,000 Associates Degree students left in Aug of 2014, and as they cut programs they "teach out" - meaning they allow the enrolled students to finish the degree they signed up for. Assuming from the GE data (and the earnings disparity) that the Associate's Degrees pretty much have to go away, the other degree declines are 21,500 over a two year period, or a CAGR of 6%. Some of that decline can be attributed to specific problematic degree programs at both the Bachelor and Advanced levels.  

But is there a healthy core? GE data suggests yes.  

The core of students from the 2012 data who were in programs with a median D/E is shown below for different levels. Basically a high percentage of Associates programs have to go away (except certain focused areas related to IT), but a lot of Bachelor Programs and most Advanced degree programs work. (In these numbers the driver is Associate/Bachelors, I have further docked the Advanced Degree programs by 30% from the end of 2014.) 

Other factors impacting enrollment include change in marketing strategy, external environment, and competition. For example, they made a recent shift away from 3rd party lead generators, who bring students who are overall less successful. It's difficult to get any precision around the relative size of these factors. But the perspective of eliminating problematic programs changes the enrollment declines   from a tsunami to just a pretty bad decline, and suggests that at some level not too far from management projections of 150,000 they may be able to find a steady state.  

Modeling Steady State?  

The drivers for total enrollment are new student enrollment and persistence rate.      

New enrollment has come down quite significantly, with the biggest declines coming when they began deemphasizing the Associates degrees. The new enrollment numbers are more sensitive to the degree splits than total enrollments, given that these students have a lower persistence rate. Quarterly persistence was 77% for Associate students and 85% for Bachelor over the 7 year period ending in Aug 2014 (again, when they stopped reporting.) Looked at another way, while Associate total enrollment peaked at 45%, Associate new enrollment peaked at 56%. Cancer surgery (i.e., eliminating the programs that do not add value for the cost) is the largest contributor to the downward trend, and the full removal of that headwind helps the trajectory.  

Still, new enrollments have continued to decline. Competition is clearly a factor here, as marketing costs per student doubled in the past five years. The extra spend on marketing has begun to flatten out, with 1-1.5% increases in the past 2 years versus mid-teens or higher previously. Spending less on marketing is another headwind to new enrollment; this is lessened (but not eliminated) as the marketing dollars per new student flatten out.  

Given management guidance of 150K total enrollment, and assuming persistence at 81%, the next four quarters would have an implied new student enrollment of just under 23,000.

Persistence has been a fairly steady data series over time. But--if new enrollment is going down because they are being more selective, then will persistence go up? They have been increasing standards and emphasizing better student acquisition channels, so this is conceivable. The most recent numbers do look a little better. However, this is difficult to prove. Persistence will be a key number to watch.  

So to frame this, where does this number need to get to? See below table which quantifies the number of new students needed each quarter to stabilize total enrollment (top line), for various persistence levels. At 81% persistence and 23K new students per quarter, total enrollment stabilizes at ~120,000. If they were able to drive persistence up to 83% and take in 30K new students (implied management case), then total enrollment pushes back up to 175,000.

Earnings algorithm  

For the US business only:  

Margins: Consolidated margins dropped from 28% in 2010 to 7% in 2015. Backing out Global, UoP EBITDA margins have come down from 34% in 2010 (peak revenue) to 17% in 2015 while revenue is down 52%, for ~50% decremental margins.  

At 150K students in a steady state, revenues fall 30% from 2015's UoP revenue of $2,148 to $1,500M, which is about $650M of revenue decline. Scoping EBITDA at different levels of decremental margin:  

Margin Thesis: fixed costs in the long term are less than the numbers have shown over the past five years. Management has done a decent job in cost cutting, but the steepness of the revenue declines have been hard to keep up with and there is a lag factor on many cost cuts. Listed below are key buckets of fixed costs over the past five years (numbers adjusted to back out the impact of Global assuming similar cost structure; calculation is the ratio of fixed cost % to total current cost % over the past five years of revenue declines.)

• Instructional/student advisory (45% of revenue, up from 33%, implied 45% fixed). Faculty are almost all adjunct and their contracts can basically be cut real time. Facilities are significantly variable over the long term, but with a time lag factor. Technology (learning management system) was essentially fixed previously, but their recent decision to outsource this essentially converts it to a variable cost. I estimate that fixed costs are ~25% of this segment; if so, the lag could take a couple years, but they ultimately get to 40% costs. This implies that decremental costs in this segment are less than 40% as lagged cuts catch up.

• Marketing, 18% of adjusted revenue, up from 12% in 2010. This is also mainly variable, but the cost escalation from competition drove it up. Should be almost all variable going forward, representing decremental costs of ~18%

• G&A, 9% up from 6%, and a good amount of this is fixed. They have some cost cutting initiatives ongoing. Assume decremental margins are a little better than 12% again because of lagged costs, 10%-11% going forward.

• Admissions advisory and uncollectible accounts: to date they have cut these costs quicker than revenue declines. Just assuming flat margins from here, this is a combined 10% of revenue.  

Summing the above would lead to decremental costs of 79% of revenue, meaning that there is opportunity to see margins increase slightly from here when enrollment stabilizes. This may not happen in F2016, as the timing on the lag is hard to figure out and the data is noisy, but this seems like a reasonable prediction for F2017 margins given stabilized enrollment around 150K.  

Management guides to 15-20% EBIT on 150K enrollment (EBITDA would be ~4% better), which would be > $285 in EBITDA. I come in shy of those levels, but the above margin analysis is how you would get close to that. In my base case, I see enrollment bottoming at 140K, recovering to the 150K level, and sustainable EBITDA being in the $150-200 range.  

In a downside case, take steady state enrollment to be 90% of the students in 5% D/E or better programs. That's 105K students and a pretty bearish case, as student results of 5% D/E matches the levels seen in a lot of good quality public univeristy programs. In that case revenue is a little ~$1,050M. Using 30% decrementals, you end up with $50M in EBITDA for UoP. The last few years capex for UoP segment has averaged $10M, and if they are indeed shrinking (along with variablizing tech expense) I don't see a reason in this scenario it would go up much. So $40-ish in EBITDA-Capex. You can pick the multiple you want to put on that, I pick 6x on the conservative side, for $240 of value in a bear case.  

Global Business  

I will spend less time here, as this is less controversial, and again focus more on the downside.    

They have a global business with 391 in 2015 revenue, with revenues in 8 countries. Consensus for 2016 is $455 (16% growth) prior to their recent German acquisition (estimated $40-50M of revenue.)   It's losing money due to investments in some newer and high growth regions, which somewhat masks the profit potential. Management has guided to 15-20% operating margins in Global over time.  

·         Largest is BPP in the UK, with a degree granting law school, plus business and exam prep. $230M in revenue, growing high single digits in constant currency.

·         Open Colleges in Australia, with 700K online students and ~$100M in revenue. Accounting is moderately complicated, basically it shows operating losses due to investment but cash flows are positive in 2016 and it longer term margins are likely comparable to other geographies.  

Comps: Kroton (Brazil) trades at >3x EV/Sales, 10x EV/EBITDA but is growing in the mid 30%'s in a fast growth Brazilian market. Recent transactions elsewhere have gotten done for >2x revenue. Apollo just paid 13.5x EBITDA for a business in a market growing closer to UK rates than to Brazil; they didn't disclose the margins but assuming it's in the neighborhood of 15% EBITDA the business does about $50M in revenue, and APOL paid a little over 2x revenue. Obviously you don't want to use what Apollo pays to mark its own business to market, but even if they overpaid by 50% it's an EV/Sales ratio of over 1. Using 1.0x forward sales as a minimum, Global is worth at least $5 a share.  

Cost basis: They bought their last 14.4% stake in Global back in 2012, for an implied value of $340. Since then they've made almost $300M more in acquisitions, so on a cost basis they paid over $6 per share for these assets, which have all grown. Again, even assuming they paid premium prices, there's pretty significant value here.  

Management case: 20% growth to end of decade, with 15-20% margins. At the low end, that's 130M in EBIT by 2019. If that were to happen, Global is certainly worth >$13 per share, and if the growth rates are continuing at that rate someone likely is willing to pay significantly more.  

Risks  

What could make the cash go away?  

Fines: the FTC issued a CID in July 2015, related to military marketing practices. The California AG has  issued a subpoena related to the same areas. A fine is a reasonably likely outcome.  

While it's difficult to get completely comfortable with anything regulatory, this doesn't seem like a complete black box to me. The FTC ramped up their scrutiny of for-profit colleges in late 2013. They have smart, competent investigators who turn over a lot of rocks. A brief list of past fines from FTC, DoE or State AG's: Corinthian Colleges $30M, Bridgepont's Ashford University $7.25M, privately held Ashworth College $11M, Career Education $10.25M, Education Management's Argosy University $3.3M. In most cases the deceptive marketing was for supplying false information about job placement rates, licensing qualifications that were in fact not met by the degree, or graduation rates.  

The headline outlier to these fines is the CFPB judgment against Corinthian ($530M, which is uncollectable given its prior bankruptcy.) Fines related to making loans are in a different category, but Apollo was not in the business of making private student loans.    

To attempt to cap the impact here, you can look at fines per student or relative to revenues or market cap. Fines per student (based on 2012 GE data) on a combined basis: Career Education $135, Ashworth $220, Bridgepoint $447, Corinthian $466. As impacted students are generally a fraction of the combined, the restitution per student can be larger: the NYS OAG estimated eligible CECO students could receive approximately $1,000 or possibly a little more in restitution, and Argosy's $3.3M fine related to just 66 students of several hundred in their doctor of psychology program, meaning a reimbursement more like $50,000 for blatant misrepresentation of their future ability to actually practice clinical psychology. Like for like, UoP's GE enrollments were 240K, so at $400 per student you end up at $96M.  

Bears on APOL could take the $1000 or even $50K number and multiply across APOL's student base and come up with an extremely high number. Why is this unlikely to be the case?  

The biggest reason in my view has to do with company and regulator motivations. Regulators are motivated to get wins; if you get a fine out of a company and can move on, that's good for press releases and your career. Companies are motivated to make problems go away, and will pay out some percentage of revenue, market cap, available cash, etc. to do that. At some point, the fine becomes high enough that the company is more motivated to fight to the death. The case drags on, the burden of proof is higher, the risk of an embarrassing public loss for the regulator goes up, the actual reimbursement to students is years away. Logic and history argue for the likelihood that APOL's eventual fine (and base case I assume that there will be one) would be in the neighborhood of past fines, and possibly larger in accordance with APOL's scale.  

$100M (~$1 per share) seems like a reasonable bad case here, with $200M being an outlier possibility.  

Capital Allocation:  

The other problem with fully valuing the cash is management's history of allocation. They bought back stock around the top of the market and increased purchases until 2012, but aren't buying back now. They trade at ~1x EV/EBITDA and are making acquisitions in new geographies at EBITDA multiples in the low teens. While the German acquisition is a legitimate business and may prove to be worth somewhere near the full multiple they paid, you unfortunately can't take cash at book value.  

I discount the cash by 25% in a base case to reflect the view that allocation is significantly suboptimal.  

Other Risks:  

Further, the company has a dual class share structure and is effectively controlled, and Chairman Peter Sperling clearly continues his father's support for management (while the family sold >$800M of stock by 2010). Management is well paid and enviably secure in their jobs despite the collapse in results and arguably a response that was way too slow; a convincing argument could be made that with proper shareholder accountability there should have been a management change well before now, but this looks unlikely to happen in the near term. Note: In late October, 2015 Peter Sperling sold 3.2M (non- voting) shares in an involuntary sale pursuant to a foreclosure. There is nothing to indicate that this leads to any changes in control or strategic direction.

Other risks flagged by shorts in the past related to the ongoing business have subsided:  

• For a while, there was a legitimate concern about APOL losing Title IV eligibility due to 90/10 rules (maximum cash revenue that can come from Title IV.) This concern has subsided as their ratio has fallen to 80% due to the aforementioned changes in the student base and the rise of corporate sponsored students.

• DoD tuition assistance does not represent a game changing risk (<1% of net revenue).  

• Concerns about accreditation were raised related to governance of the UoP, but such concerns were resolved

I've argued that you have a margin of safety from the (discounted) cash and the Global business. To really get killed from here, you'd need some combination of Global falling apart and fines that take away all (or more) of the cash.  

Conclusion  

APOL is clearly a hated name, trading at low multiples for very good reasons. The above discussion by no means caps all the risks: regulators can act unpredictably; a "healthy" core of students who were getting value from the program in 2012 does not ipso facto mean that there is a healthy core at that  level in a different environment in 2015 and beyond; investing in a controlled company with a suboptimal capital allocation is not desirable. Still, the discount in my view is significantly greater than the negatives. Position sizing should obviously reflect the above risks.        

Catalyst

Biggest catalyst is just cheapness relative to cash flows.  Stock could move based on:

- enrollment numbers stabilize

- global growth continues and turns earnings positive, removing potential doubts on global value

- clarity on regulatory activity (though full clarity here seems unlikely in the near term as there are mulitple moving parts)

All of this takes time, and I do not expect it to play out sooner than 4-8 quarters from now.