The Value of This Portfolio

I've long admired Motley Fool Hidden Gems recommendation Portfolio Recovery Associates (NASDAQ:PRAA) -- last summer, I even went so far as to call it "A Stock Lynch Would Love." PRA buys up blocks, or "portfolios," of uncollected debt from credit card companies and the like for pennies on the dollar, then typically collects several times more than it paid.

But while I've always thought the Portfolio Recovery story was excellent, I didn't have a great handle on valuation -- after all, a great story does not necessarily make a great investment. The accounting was convoluted and appeared overly susceptible to management "decision-making."

For example, revenue recognition is tied to management assumptions of portfolio collections. Using different assumptions, for each $1,000 collected, Portfolio Recovery recognizes revenue of approximately $700. Competitor Asset Acceptance Capital (NASDAQ:AACC) recognizes and reports revenue of approximately $800. At a glance, PRA might appear to have slower revenue growth, but it's a trick of financial reporting. The disparity doesn't change the fact that both companies collected the same amount. Similar problems crop up with reported earnings and assets. These problems mean that conventional financial analysis metrics and definitions of free cash flow or owner earnings fail to accurately portray PRA's valuation.

Take another look
What if we simply ignored the distortions of accrual accounting? If we focus solely on cash going in and out, we can model PRA as a giant time-value-of-money problem. (Stop groaning out there!)

Skipping ahead to my final model:

So, then, to finish up, simply forecast each item over the life of the model, calculate free cash flows, discount free cash flows back to the present, add cash, subtract debt and other claims against value, and ... boom! We have value per share!

Easier said than done. The lifeblood of Portfolio Recovery Associates is new portfolio purchases. So at what growth rate should we model future receivables purchases? Debt purchases historically have been rather lumpy; management has said that all new purchases must meet their internal rate of return targets.

How about bankruptcy-purchases growth? Management is now comfortable with the collection aspects of this debt class -- enough to be buying heavily in the fourth quarter of 2005, but the history here is very limited.

The past five years of purchases (in millions) have been as follows.

How should we model future collections performance? Management targets a 250% to 300% return on invested funds over a six- to seven-year portfolio life. But all portfolio purchases are not created equal. Portfolios purchased in 2000-2002 have blown past the upper target. Conversely, a portfolio purchased in today's tighter pricing environment is likely to come in at the low end of the target. Changes in the accounting rules in 2005 also give management incentive to estimate the collection profiles of new purchases initially on the low side, so as to avoid reporting future impairments. (Not suggesting anything untoward, but viewing the company on a cash-collections basis eliminates this concern)

Fortunately, we can use the history of collections to help us in our forecast.

Cumulative cash collections-to-purchase price --
Cash collection period (years) since purchase

There are other problems to address. How do we decide on a cash tax rate? What forecast period should we use? A typical discounted cash flow (DCF) analysis is broken into two parts -- an explicit forecast period where we expect outsized growth and a terminal value period where we expect growth in free cash flows to grow slowly, just keeping pace with inflation. But not for Portfolio Recovery Associates. We cannot simply move from the high-growth period -- say, 20% annually -- to the low-growth period -- say, 3% annually.

Why? Imagine a portfolio purchased in the final quarter of the explicit forecast period. It will have an expected collection curve that will extend well into the terminal period; indeed, it will generate the vast majority of its collections in that terminal period. Using a conventional two-stage model unfairly cuts off value from late-purchased portfolios. The solution is to use a three-stage model: an explicit forecast period; an interim period where collections on late-purchased portfolios continue apace, but new portfolio purchases slow to the terminal rate; and finally, the true terminal period.

So what's the point?
The table below shows my assumed model inputs. I believe they're reasonable, but I am open to being called foolish (with a small "f").

Using these inputs, and incorporating as much information as is currently available from the just-released first quarter (the 10-Q is not yet filed), my fair value of Portfolio Recovery Associates is $56.

But a DCF analysis is heavily dependent on the input assumptions. Varying these assumptions gives an idea of the robustness of the model, but it also shows that you can pretty much justify whatever value you want. I believe I've been conservative with my assumptions, so PRA's true value may be higher.

On the PRA message board, I'll present full calculation details, as well as some discussion on model sensitivity. Hopefully, this number is enough to entice you to the boards!

This article is adapted from a past column in Hidden Gems Daily.

For the original buy report on Portfolio Recovery, as well as Jim's full calculation details and a discussion on model sensitivity on the Hidden Gems discussion boards, click here for a free 30-day trial to Hidden Gems.

Jim Gillies owns shares of Portfolio Recovery Associates (anyone surprised?). The Motley Fool has a strict disclosure policy.