Portfolio Recovery Associates, Inc. (“PRA” or the “Company”) has more than doubled in value in the last year as collections and earnings have repeatedly surpassed expectations.  But the market has failed to recognize the low quality of PRA’s earnings. Seventy percent of the Company’s operating earnings derive from portfolios acquired three to four years ago.  In the intervening years, portfolio prices have risen dramatically and returns on newly acquired portfolios have declined commensurately. These older, highly profitable portfolios will “burn out” over the next year or so and will be replaced with newer portfolios generating much lower returns. PRA’s earnings momentum will decelerate if not reverse entirely, returns on investment will decline and PRA’s stock valuation, at 3.5x book value (4.2x TBV), will appear increasingly stretched.

PRA business

PRA operates in two related, but distinct businesses: (1) the acquisition and collection of defaulted receivables, primarily credit card receivables (“Core”), and (2) the acquisition and collection of receivables of individuals in or just out of bankruptcy proceedings (“Bankrupt”).[1]  Although PRA treats the Core and Bankrupt businesses as one, the characteristics of these two businesses are quite different. The Core business is characterized by relatively high collection multiples (i.e. expected total collections as a multiple of purchase price), high collection costs and long collection periods.  By contrast, the Bankrupt business is characterized by low collection multiples, low collection costs and short collection periods. Since the financial crisis, both of these businesses have significantly outperformed market, as well as management, expectations.

PRA accounting

The accounting for PRA is opaque.  The accounting treatment is governed by ASC 310-30 and works as follows. Management projects future cash flows for each acquired portfolio.  Management next computes the effective yield that fully amortizes the portfolio based on the projections (the “IRR”). Thus, on the balance sheet is an asset, Finance receivables, net, that initially is recorded at its purchase price. Income from the receivable is accrued at the IRR, which increases the receivable, which is then reduced by the collections received.

Here is an illustration:

PRA acquires a portfolio of defaulted credit card receivables for $100 and calculates an IRR of 35% (ignoring all collection costs and taxes) for the portfolio.  Collections in the first year are $50. 

Income: $35 (35% x $100)

Receivable balance at year end: $85 ($100 + $35 - $50)

Thus, income is independent of actual collections (at least in the short term).  Regardless of actual collections in the first year, PRA will recognize finance receivables income of $35.  This example highlights the importance of the collection multiple on GAAP earnings.  The higher the expected collection multiple, the higher will be the IRR.  The higher the IRR, the higher will be the accrual rate.  And a higher accrual rate means that PRA recognizes as income a higher percentage of its collections as revenues.  Since operating expenses are a function of collections, margins are much higher on high collection multiple portfolios.

To ensure that collections and income remain in accord, management makes a quarterly determination as to whether future collections will differ from expected projections.  If such a difference is found, the accounting is asymmetric. If collections fall short of initial projections, PRA must take an immediate charge equal to the difference between the present value of the new projected cash flows, discounted at the original IRR, and the receivable balance for that portfolio. If collections exceed initial projections, there is no increase in the value of the receivable balance.  Instead, management calculates a new IRR such that the present value of the new projections equals the receivable balance.  Thus, income from an outperforming portfolio will be higher in future periods as the income recognition rate is higher. The effects of this accounting treatment are twofold: (1) Management conservatively estimates initial cash flows so that the portfolios have a performance cushion before any charges must be taken, and (2) The income effects of acquiring outperforming portfolios can last for years, which means that PRA’s GAAP return on investment can far exceed the economic returns available on newly acquired portfolios.

The Economics of Debt Collections

The economics of PRA’s business are relatively straightforward.  There are only three variables: portfolio cost, collections and their timing, and costs to collect.  Let’s analyze these in reverse order.  Because PRA fails to break out costs between the Core and Bankrupt portfolios, it is necessary to estimate indirectly the costs to collect the Core portfolios.  As noted above, the two portfolios have vastly different characteristics.  For the Core portfolios, the collection process is an active one, with collections prompted by litigation, or its threat, as well as contact from call centers.  For the Bankrupt portfolios, PRA does little other than cash the check.  The payments are received by the bankruptcy trustee and, pursuant to a Chapter 13 plan, disbursed to the various creditors.  Creditors, like PRA, are in fact bared by law from making collection efforts. The chart below attempts to calculate the cost to collect PRA’s Core portfolio.  I have assumed that costs are 5% of the amounts collected for the Bankrupt portfolios.

What is immediately obvious is that PRA’s cost to collect the Core portfolios has not really come down.  Any perceived decline in total costs as a % of collections is merely an artifact of the methodology.  By combining Core collections with low cost to collect Bankrupt collections (which have grown at a faster rate than Core collections), it appears that the cost to collect has decreased.  In fact, costs have held fairly steady during the last few years. 

The other two variables, portfolio costs and total collections are closely related.  Ceteris paribus, the cheaper one can buy a portfolio, the higher will be the collection multiple (i.e. the amount ultimately collected as a multiple of the purchase price).  Of course, purchase price is not the only factor affecting collection multiples.  Changing economic conditions of the Core and Bankrupt accounts has a significant effect.  Improved collection performance from better data mining should also increase multiples.    

The chart below details the estimated collection multiples for the Core vintages for the last 11 years.

The pattern is clear.  Portfolios acquired in the early years of an economic recovery outperform.  Portfolios acquired in periods of economic stability underperform.  The reasons are not hard to discern.  During and immediately after an economic downturn, capital is scarce, defaulted receivables are plentiful and expected collection forecasts are conservative.  So prices are low and collections exceed forecasts.  As the economy improves, buyers become more confident and the amount of defaulted receivables declines.  Prices rise and buyers’ elevated forecasts are more difficult to achieve.  The chart also makes clear that outperforming portfolios outperform from the start.  The first year collections for outperforming years significantly exceed those from underperforming years.[2]  The unexceptional first year collections of the 2012 vintage imply that 2012 will not generate significant upward revisions.

What about the Bankrupt portfolios?  PRA’s history with Bankrupt portfolios is briefer so there is less data to evaluate.  Despite that, the outperformance in the aftermath of the financial crisis is even more pronounced. 

In 2009 and 2010, the portfolios traded cheap as evidenced by the estimated collection multiples, which were at all-time highs.  Despite the optimism built into the original projections, both of these vintages significantly outperformed.  And the outperformance is stunning.  For the 2009 vintage, PRA’s initial estimate of cash flows generated an annualized return of 23% (after-tax).  Its current estimate of cash flows generates an annualized return of 35% (after-tax).  Now compare that to annualized returns of its more recent vintages.  For Bankrupt portfolios acquired in 2012 and during the first quarter of 2013, the estimated annualized return is 7% (after-tax).  Not too shabby, but down 80% from the peak. The reason for declining returns is obvious.  The prices that PRA is paying for Bankrupt claims have almost tripled since 2009 whereas the ultimate collectability of these claims has increased only marginally.[3]

What about the returns on the Core portfolios? For 2009 and 2010 vintages, the estimated collection multiple is 326.  Fully loaded for all costs and taxes, that generates an annualized return of 12%.  The Core portfolios acquired in the first quarter of 2013 have a collection multiple of 211, the lowest ever.  The annualized return for this vintage, calculated on the same basis, is only 1%.  Of course, the marginal annualized return is likely to be higher because my methodology loads all costs (including G&A and prepaid legal).  It is also likely that this multiple will increase over time.  The conclusion is, however, inescapable.  At current prices, returns on the Core portfolio have collapsed.

If returns have collapsed, why are earnings at record levels?

To paraphrase James Carville, it’s the accounting, stupid.  PRA’s GAAP financials do not reflect its current economic earnings, but rather the earnings of the past.  For example, in Q1 2013, PRA reported income from finance receivables (i.e. revenues) of $154.8 million and operating income of $65.9 million.  Fifty four percent of the income from finance receivables is from portfolios purchased in 2010 or before.  Even more revealing, almost 70% of the operating income in Q1 2013 came from portfolios acquired in 2010 or before.  Indeed, I estimate that half of Q1 2013 operating income came from just two vintages of Bankrupt portfolios: 2009 and 2010. [4]  Of what relevance in trying to understand PRA’s economic model is the performance of portfolios acquired three and four years ago?  Yet these portfolios dominate PRA’s financials.

The perverse effects of the accounting treatment are now apparent.  In an effort to prevent management from manipulating earnings by writing up outperforming portfolios (and booking the gains through income), the accounting rules have injected another distortion.  GAAP income does not measure the current profitability of PRA.  Instead, it is little more than a record of portfolio purchases made years ago.

When will PRA begin to show cracks?

The timing is difficult.  PRA will continue to benefit from the outperformance of the 2009 and 2010 portfolios, but their effects will diminish.[5]  But the more recently acquired portfolios are unlikely to significantly outperform estimates. Let’s examine the types of receivable PRA has been buying and the prices they are paying.  PRA (and the industry) categorize Core receivables into one of four categories: Fresh, Primary, Secondary and Tertiary.  Fresh receivables are recently charged off receivables (usually 120 to 180 days past due) on which no attempts to collect have been made.  Primary are six to twelve months old and the original creditor has hired one third party collector to try to recover. Secondary are twelve to eighteen months old and the original creditor has hired two collectors. Tertiary are more than eighteen months old and multiple collectors have been hired.  PRA has provided a breakdown of dollars spent and average prices paid by category for the last nine quarters (i.e. since Q1 2011).  During this period, almost half of the dollars spent have purchased Fresh receivables and another 26% have purchased Primary portfolios.  The chart below details the market prices/prices paid for charged-off credit card debt. 

The chart highlights a complete round trip, from peak prices in 2007 to a nadir in 2009 and a return to peak prices in 2013.  Bear in mind that for 2007 Core paper, PRA initially estimated a collection multiple of 227 and the actual multiple has turned out to be 225.[6]  For 2012 and Q1 2013, PRA has estimated collection multiples of 226 and 211, respectively.  Current multiples imply total recoveries of up to 30 cents on the dollar for Fresh portfolios acquired in 2011 and 2012 portfolios. In a world where the “vast majority” of charged off consumers never make any payments, it is difficult to envision average recoveries significantly greater than 30%.  Thus, significant upward revision to collection multiples for recent vintages do not appear likely.

The price escalation of Bankrupt claims has been even more dramatic.  In 2009, PRA paid 5% of face for Bankrupt claims.  In Q1 2013, PRA paid 20% of face.  To be sure, these claims included some secured claims, which is a new business for PRA, and generate higher collections (and thus higher purchase prices).  In Q4 2012, PRA paid 11.7% of face for Bankrupt claims, which included no secured claims. Another framework for analysis is to compare the ultimate collections for 2009 and 2012 Bankrupt paper.  For 2009 paper, PRA is forecasting total collections of 14% of par.  For 2012 paper, PRA is forecasting total collections of 17% of par. Collections for Bankrupt portfolios are thus fairly consistent over time. Because the purchase price has risen so dramatically, returns have collapsed. It bears repeating, the Bankrupt claims collections cannot be enhanced by improved collection techniques.  All PRA does is buy the paper and receive payments from the bankruptcy trustee.  In a competitive market and a zero interest rate world, returns on Bankrupt claims are likely to be in the mid single digits.

Just how competitive is the market for Core and Bankrupt claims?  One data point is the recent merger of two of PRA’s publicly traded competitors: Encore Capital Group (ECPG) and Asset Acceptance Corp (AACC).  ECPG is acquiring AACC for a mixture of cash and stock.  The acquisition is really nothing more than an acquisition of a large portfolio by ECPG, as there is little, if any, franchise value at AACC.  The price ECPG is paying for AACC reflects a collection multiple of 2.1x for AACC’s portfolio.[7]  It is revealing that ECPG management believes it better to lock in a large acquisition at a collection multiple of 2.1x, which is a historically low level. ECPG management has announced that it will largely exit the market for new charge-offs and bankrupt claims for the balance of 2013.

Normalized Earnings

What are normalized earnings for PRA?  Let’s assume that the 2009 and 2010 Bankrupt portfolios did not generate outsized collections, but generated collections consistent with PRA management’s estimates of recently acquired portfolios (136% collection multiple for Bankrupt portfolios acquired during the last five quarters). Let’s leave the other outperforming Core portfolios unaffected.  On this basis, I calculate that 2012 EPS declines from $7.39/share to $3.85/share and that Q1 2013 EPS declines from $2.26/share to $1.16/share.[8]  Thus, PRA’s normalized PE ratio approaches 40x. 

Side by Side Comparison

PRA has one significant public comparable: Encore Capital Group.  On the most important metrics, the two companies are very similar.  ECPG’s record on driving down collection costs is superior to PRA whereas PRA’s ability to drive higher collection multiples is superior to ECPG.

These two effects largely cancel out, which means the two companies earn very similar returns on their collections.  Whereas PRA is valued at 19x earnings and 3.5x book value, ECPG is valued at only 10x earnings and 1.6x book value.  Why the difference?  ECPG has only recently become active in the Bankrupt claims business.  As a result, ECPG has not benefited from the outperforming vintages of 2009 and 2010.  In all other respects, these two companies are quite comparable, except that PRA is valued at a 100% premium. In a world where newly acquired Bankrupt portfolios generate no better than mid-single digit returns, PRA should be valued consistently with ECPG.

PRA also faces tax risk.  The Company has utilized the cost recovery method in determining its taxes.  Under the cost recovery method, collections on receivables are applied first to principal to reduce the finance receivables to zero before any taxable income is recognized.  Following audit, the IRS has asserted that this method does not accurately reflect taxable income and issued a Notice of Deficiency.  PRA has appealed to the US Tax Court.  PRA has a deferred tax liability associated with the cost recovery of $190mm.  With interest and penalties, the amount at issue is likely to exceed $250mm.  As a point of reference, ECPG and another competitor, Asta Funding, did not utilize the cost recovery method, whereas a third competitor, Asset Acceptance utilized cost recovery.  A loss on appeal, coupled with aggressive purchases of receivables in the coming years, will stress PRA’s balance sheet.

Risk

Despite paying extremely high prices for Core and Bankrupt claims, PRA manages to generate collections significantly above historical levels and earn commensurate returns.