Description

Market data 10/27/2017 via Bloomberg. 2017 metrics TTM GAAP

52-Week Range: $23.15 - $42.70

5-Year Avg ROE: 19.75%

Estimated Remaining Collections: $5,325m        

Year-end 2019 ‘normalized’ Target Price: $60

Year-end 2019 ‘no-growth’ Target Price: $43

Catalyst: Declining amortization rates.  Every dollar allocated to or freed up from amortization charges drops directly to pre-tax earnings, creating optics of tremendous operating leverage and growth of GAAP results.

Specifically, I believe amortization rates are artificially high currently as management is booking purchases at overly conservative assumptions to save on cash taxes paid in 2017 while waiting for clarity on corporate tax reform and/or free cash flow generation to catch up to the improved purchasing environment in the United States.  I expect amortization rates to begin to fall in Q1 2018, or as soon as clarity on tax reform is reached, and to be significantly lower by 2019-2020.  This will create growth in reported GAAP numbers in excess of fundamental cash collection growth.

Thesis: PRAA is a cyclical company with complicated accounting currently trading at a no-growth valuation.  The fundamental economic cycle for most of the business bottomed in 2015, but we are being offered the same investment opportunity today.  A combination of government regulation, lag in company results, and conservative management have masked improvements in company results that will result, at some point, in improved reported results and significant growth in EPS.  As a new oligopoly in the US due to CFPB regulation, PRAA stands to benefit from an up cycle in the industry.  No one can predict the specific date when results will improve due the difficulty of forecasting amortization rates, creating opportunity to invest ahead of others if you have a longer time horizon or a trigger if you prefer to wait for a trend to start.

Revenue Model/Accounting & Terminology

This is available in filings, but typically generates the most questions, so feel free to skip forward if not needed.  I’d particularly direct your attention to slides 16-17 of the September 2017 investor presentation for more detail.

First and foremost:  GAAP accounting obscures the economic underpinnings of the business.  Revenue is a metric of gross profitability for PRAA, materially understating cash generated by the company.  A typical revenue cycle looks like this:

1. Purchase charged-off debt and record at purchase price on the balance sheet as Net Finance Receivable (NFR)

2. Estimate the cash to be collected over an assumed 10-year life.  This is management's primary location to make assumptions on how much will be collected and in what year.  Total dollars collected peak in years 2-3 of collections and trail off much like an oil well decline curve over time.  The sum of all estimated cash collections is Estimated Total Collections (ETC).  ETC / Original purchase price (NFRs) = Purchase Price Multiple. The portion of ETC not yet collected or written off is Estimated Remaining Collections (ERC).  These items are disclosed as supplementary information in 10-K and 10-Q reports on a vintage year basis

3. Start collections.  Perform an IRR calculation on the NFR purchase price and estimated cash flows from ERC over the remaining life of the debt.  This IRR is the Gross Yield (IRR) of the vintage.  Gross yield and cost to collect are the primary purchasing decisions for management, and I believe 30-40% IRRs are management’s minimum hurdle rate for purchasing of credit card portfolios.  In periods of limited supply (2015) they may reach to 25%.

4. Report results via GAAP.  Period-start NFR balance * Gross Yield (IRR) = Revenue recognized for the period.  Note that revenue from existing portfolios is determined at the beginning of the reporting period by the Gross Yield, itself derived from cash collection estimates.  Cash Collections less Revenue = Amortization.  Most amortization is “cash applied to NFR balance” but also includes any Allowances or write-offs of NFRs deemed uncollectable during the period.  Cash Collections less Amortization less Allowances = Revenue reported on GAAP statements

5. Readjust ERC cash collection assumptions, impacting ETC and Purchase Price Multiple.  Calculate a new Gross Yield (IRR) using the remaining unamortized NFR balance and estimated ERC cash flows.  Rinse and repeat.

No-Growth Valuation / Current Situation

Several factors over the last few years have impacted each of the four reported segments of PRAA and left the business with no- or negative-growth rates on all important metrics.  The market has not responded favorably.

Americas Core – 56% of cash collections FY2016.  Growth in small Brazil & Canada operations has not offset decline of domestic credit card collections.  Tight credit after the financial crisis limited supply of new charged-off debt, hurting PRAA’s profitability.  Low supply and/or high competition lead to higher purchase prices and lower returns.  CFPB regulation/uncertainty further limited supply as 3 major sellers of charged-off debt (Wells Fargo, Chase, Bank of America) suspended debt sales, representing ~30-40% of total annual charged-off credit card debt.  Supply bottomed in 2015 per my calculations.  Large allowances (write-offs) of NFRs were recorded in 2015-2016 for debt that was deemed no longer collectable under new CFPB regulations. Cash collections were (1%) YoY in FY2016

Americas Insolvency – 17% of cash collections FY2016.  The OCC updated/muddled regulations on the sale of bankruptcy auto loan debt in 2014.  PRAA essentially stopped all bankruptcy purchasing due to low supply/high prices in 2H 2014 resulting from OCC changes through Q1 2017.  In FY2013, this segment represented 41% of total cash collections.  Cash collections were (27%) YoY in FY2016.

Europe Core – 26% of cash collections FY2016.  The 2014 purchase of Aktiv Kapital came with known slower portfolio recovery rates than US collections (ie: lower gross yields), and the takeout premium of the purchase lowered the expected return on Aktiv’s sizable existing Europe portfolio.  European pricing environment has worsened since 2014, and collections here have been unable to do more than offset the decline of Americas Insolvency business.  Cash collections grew 14% YoY in FY2016.

Europe Insolvency – 1% of cash collections FY2016.  Fast growing but too small to move the needle.  Cash collections were up 99% YoY in FY2016.

Due to the lack of growth, PRAA trades near its 1.0x EV/Net ERC valuation.  This is not a run-off or liquidation valuation, but is akin to a bank’s 1.0x P/B metric, assuming PRAA is only able to purchase enough new debt to replace the ERC collected each year with no improvement in profitability.  Net ERC is calculated as total ERC less the cash cost to collect on the debt.  I use each quarter’s cash operating costs / cash collections as the cost to collect, but this metric has consistently been near 40% of cash collections historically.  Q2 2017’s data of $5.3b ERC and $1.9b net debt leads to a 1.0x EV/Net ERC stock valuation of $28.77.  This excludes some small lines of business, portfolios not on accrual accounting, investments, and other minor items.

Since Q3 2015 when revenue growth first went negative, the playbook has been for a weak earnings report to be sold down to this 1x EV/Net ERC area, from which the stock typically grinds higher into the next earnings report.  While not a “floor” in the stock, this certainly appears to be where the value bid can be found.  Q1 2017 was the first truly good earnings report since 2014, followed up by a clunker in Q2 with disclosures that the market assumed to be a no-growth guide.  (Note: management does not give guidance)

Long-Run Valuation Model

When results are decent (or mid-cycle), I believe the market primarily values PRAA on a P/E basis due to “too hard” issues, with a historical range of 10x-20x (ttm) depending on current outlook and observed revenue growth.

The goal of the following exercise is to give a rough forecast of results in a normalized environment (primarily shown via amortization rate, discussed later).  I believe a full normalization is reasonable to expect by 2019-2020, but to facilitate comparison with published consensus estimates, I am using my 2019 estimates for valuation.  Note this may be optimistic for a full year of normalized results, but the trend should be extraordinarily clear in business results by this time.

1. Run-off of existing portfolio from year-end 2016.  Assume zero new purchases, zero growth in ETC for owned portfolios, ~6% of current ERC being lost to allowances, 40% of cash collections cost to collect (60% cash operating margin), no new borrowing, 4.5% interest cost (~3.8% actual), and 40% tax rate:

Results:

2019 Estimated Remaining Collections (ERC): $1,813m

2019 Revenue: $472m

2019 Portfolio Amortization Rate: 47%

2019 EPS: $0.61

2. Europe continues with no growth and subdued purchasing.  Assume core purchases at a 165% multiple, insolvency purchases at a 130% multiple (2016 actuals), and no future growth in ETC.  That is, ignore all potential for growth in this segment, which implies ~$300m annual core purchases and ~$40m annual insolvency purchases.  FY2016 purchasing was $352m Europe Core and $43m Europe Insolvency.

Results:

Assumed Annual Purchasing: ~$340m

2019 Incremental ERC: +$1,301m

2019 Incremental Revenue: +$87m

2019 Portfolio Amortization Rate: 49%

2019 Incremental EPS: +$0.09

3. ETC Growth for existing Americas Core portfolio.  Historically, purchase price multiples have expanded over time, especially for credit card portfolios.  This speaks to management’s conservative estimates and their ability to successfully collect on debt.  The 2008 vintage portfolio was booked at 220% purchase price multiple and is currently at 225%, which is by far the worst performing portfolio on ETC growth over time.  Since the 2002 IPO, the average starting multiple in this segment has been 228%, with the average multiple expanding to 297% by year 5 of collections, with further gains through the 10-year estimated life.  Historically, ETCs will grow 5-10% per year for the first 4-5 years of collection, and grow ~2% per year for the remainder of the assumed 10-year life.  See the purchase price multiple supplemental disclosures in 10-K/Qs for trends and more detail.  Note that the 2015 and 2016 CFPB-inspired allowance write-offs removed an estimated $63m and $228m of ERC, respectively, effectively hiding the growth of ETC in remaining “good” debt.  On a 1x EV/Net ERC basis, these allowances removed $1.40 and $5.05 from share valuation, respectively.  I believe the industry is past all CFPB-related allowance charges as formal rules were published in 2016.  The chart below illustrates historical ETC growth for the Americas Core portfolio, by vintage.  Data includes effect of allowance charges (write offs of NFRs) which is visible as downticks in most recent year data for vintages.  

Assuming: 2015 vintage performs similarly to 2008 vintage with little to no ETC growth.  All other 2010-2014 vintages perform as follows: ETC growth of 8%, 10%, 6%, 4% for years 2-5 of collection, respectively, and 1% growth for years 6-10.  The 2016 vintage performs at 2010-2014 assumptions less 100bps in years 2-5.  No new purchasing.  No modeling of vintages prior to 2010.

Results:

10-Year ending purchase price multiples of 223% for 2015 vintage, 260-270% for others

2019 Incremental ERC: +$275m

2019 Incremental Revenue: +$61m

2019 Portfolio Amortization Rate: 46%

2019 Incremental EPS: +$0.47

4. Normalization of US Credit Card market purchasing.  Supply of charged-off debt bottomed in 2015.  See chart below, but note various unadjusted data giving longer historical perspective is available on FRED such as https://fred.stlouisfed.org/graph/?g=fccr.  

American Express and Discover do not sell debt, and Bank of America, Wells Fargo, and Chase have remained sidelined since ~2014 due to regulation changes/uncertainty but remain in contact with PRAA per management.  Using FFEIC call reports, American Express and Discover represent 8-10% of the annual supply of charged-off credit card debt, while the sidelined sellers account for roughly 30-35% of the remaining supply of purchasable debt.  Historical expansions of charged-off debt typically last 2-4 years, but we are coming off a prolonged downturn in supply which may result in an abnormal expansion.

Assuming: 4% long-run credit card balance growth and a recovery to a long-run 4% charge-off rate of credit card debt by 2019, with current sidelined and non-sellers staying out of the market and maintaining current ‘market share’ of charged-off debt, 2022 estimated US supply will grow ~54% from year-end 2016.  Obviously, there will be cycles above and below the trend rate.  From today, an economic boom is likely to be fueled by growth in consumer credit balances while an economic bust is likely to have negative growth in balances but a significant move higher in charge-off rate.  PRAA supply should benefit from either situation due to the historical lows in 2015 as one of essentially two remaining US operators due to oligopoly situation created by CFPB regulation in 2014-2016.

Also assuming PRAA maintains 35% market share of purchasing (2013-2015 actual: 36.8%, 38.2%, 32.6%) and is able to purchase debt at ~$0.08 per dollar of face value (2014 actual, historical lower, 2015 ~$0.105/dollar face), PRAA purchase volume in dollar terms of US credit card debt will grow ~69% by 2022 from 2016 levels.  (Note: charged-off debt is often sold or collected multiple times before reaching PRAA, who does not resell purchased debt.  This can obscure market share calculations and causes results to lag purchasing conditions)

Furthermore, assuming PRAA can purchase this debt at a 200% purchase price multiple (2012-2016 actual average 209%) and have ETC grow at normal rates to a 280% 10-year exit purchase price multiple (historically conservative, but likely reasonable for increased market share), and a 2% annual allowance rate, I see the following contribution to results:

Results:

2019 Incremental ERC: +$2,349m

2019 Incremental Revenue: +$452m

2019 Portfolio Amortization Rate: 42%

2019 Incremental EPS: +$2.42

5. Resumption of Americas Insolvency debt purchasing.  This segment is primarily domestic auto loans having court-ordered bankruptcy settlements.  Purchase prices and amortization rates are much higher, but courts handle collection and PRAA is essentially required to only file appropriate paperwork for the collections to be passed through to the company.  After years of essentially no purchasing, we have seen this segment return in 2017.  A few ways to quantify: $97m in domestic auto deficiency purchasing YTD 2017 vs $6m and $7m for 2016 and 2015, respectively.  Also, $167m total domestic insolvency purchasing (auto + credit card) YTD 2017 vs $69m full year 2016 and $65m for full year 2015.

Market size data is less readily available.  Average auto loans outstanding and growth rate is available here https://fred.stlouisfed.org/graph/?g=ffOs.  Capital One holds approximately 4.75% of domestic auto loan market, and has historical annual charge-off rates of roughly 1.7% per Bloomberg.  If we assume this rate is indicative of the entire domestic balance of $1.09 trillion outstanding in April 2017, and 20% of that number is subject to bankruptcy proceedings (frankly, a guess), and the average purchase price is approximately $0.30 per dollar of face value (PRAA average, 2012-2014), there is roughly $1.1b (purchase price) of supply available per year.  This is a very rough estimate.

We don’t know the cost to collect for this debt, but assuming it is 20% of cash collected (half of call center collection cost), a purchase price multiple of 127% (slightly below 2012-2014 average) and ETC growth of a minimal 2% annually in years 2-6 of each vintage, every $100m of insolvency purchasing contributes an incremental $0.04, $0.08, $0.05, $0.03, $0.02 in EPS per year in collection years 1-5, respectively per my estimates.

Assuming PRAA can purchase $250m of insolvency debt per year going forward leads to the following results.  (Insolvency purchasing has been as high as 50% of total purchasing in 2012; a similar rate today would be ~$500m/year purchasing)

Results:

2019 Incremental ERC: +$618m

2019 Incremental Revenue: +$69m

2019 Portfolio Amortization Rate: 44%

2019 Incremental EPS: +$0.37

Totals for items 1-5 above:

2019 Revenue: $1,138m

2019 Portfolio Amortization Rate: 43.7%

2019 EPS: $3.96

2019 ERC: $6,356m

2019 ROE: 19%

Year-end 2019 Stock @ 15.0x ttm P/E: $59.40

Year-end 2019 Stock @ 1.0x EV/Net ERC (no change in net borrowing): $43.15

Actuals for comparison:

2016 Revenue: $831m

2016 Amortization Rate (ex-allowances): 43.5%

2016 EPS: $1.83 (Q4 kitchen sick)

2016 ERC: $5,048m

2017 YTD Revenue: $407m

2017 YTD Amortization Rate (ex-allowances): 48%

2017 YTD adj EPS: $0.79

Q2 2017 ERC: $5,325m

Where this is wrong:

• Ignores all YTD 2017 actual results.  This is an exercise focused on normalized long-term results.

• My cash collection model is slightly high in first year of vintage collections, but is close to historical reported gross yield behavior

• My amortization rates are low in the first 1-2 years of collection vs management history.  This overstates near term results but understates long term results.  The economics over the 10-year life are the key issue, the difference in results being a timing of recognition issue

• It remains possible CFPB-inspired industry changes prevent future growth of ETC estimates, at which point my thesis would be broken.

• Europe purchasing may disappoint long-term.  Segment is most likely to be ‘underfunded’ near term during cash crunch situation.  No great loss given my incremental EPS estimates

• Q2 2017 YTD purchasing vs forecast:

Americas Core: $260m vs $495m forecast (usually slightly higher in 1H)

Americas Insolvency: $167m vs $250m forecast

Europe Core: $82m vs $300m forecast (2H higher, but clearly pulling back on irrational pricing)

Europe Insolvency: $14m vs $43m forecast

Amortization and Catalyst

Let me be perfectly clear: at some point, an investment in PRAA is a question of trusting management: that they’ll buy good debt, estimate conservatively, and raise those estimates in future years, thereby increasing gross yields, and thus create significantly higher growth rates on GAAP numbers.

The path to significantly higher EPS is reasonably easy, if beyond the typical sell-side publicized forecast horizon.  The reason everyone struggles to predict when results will turn is the amortization rate of cash collections. To repeat from the accounting section: every dollar charged to or freed from amortization drops directly to pre-tax earnings.  This is the source of ‘operating leverage’ for this company.  Portfolio assumptions evolve over time in an attempt to amortize NFR balance to $0 at the 10-year assumed life.  Low cash collection assumptions, or simply slow (pushed to out year) collection assumptions lower gross yields, resulting in lowered revenue.

If revenue is too high (low amortization) we should worry cash collections are underperforming assumptions and watch for future allowance charges.  If revenue is too low (high amortization) management is either paying far too much to purchase debt (bad pricing) or cash collections are outperforming assumptions, resulting in over-amortization of NFR balances, and we should watch for future improvements in gross yield to recognize more revenue.

I believe PRAA is currently over-amortizing (estimating collections in such a way to lower gross yields) its US portfolios to: 1) be conservative on estimates; and 2) minimize its tax burden in the short term. Here’s what we know, or management has stated:

• The supply environment is improving, pricing is improving, and regulatory overhangs have cleared. The industry has reset following 7 years of shrinking supply and increased regulation.

• PRAA has settled with IRS (pre-2012 tax dispute), will pay $240m deferred taxes over 4 years (no fees or interest) starting Q2 2017

• With the IRS settlement, PRAA has a new tax agreement where taxes paid will mimic GAAP tax expense on the income statement, which ran at 43.7% in Q2. Call it 40% of EBT, starting Q2 2017.

• Adjusted Debt/Equity is near 2x (1.93x) which is the level where management is comfortable. The company is not tapped out, but would rather grow without significant additional leverage. PRAA has $500m in untapped debt facility in the US for additional purchasing if needed.  No debt maturities until 2020, average rate ~3.8% with recent issuance.

• It appears that there is a bit of a cash crunch in 2017, where PRAA probably can't buy as much new debt as they’d like from only the cash generated from collections

• Management compensation is based on 3-year rolling ROEs and total shareholder return, so they should want to return to growth and buy as much profitable debt as possible.

• Management is selectively buying debt because they’re at collector capacity, which one would think would result in purchasing only the best debt available

• The company hiring a ton of domestic collectors (400 in Q1 2017, 900+ last 9 months.  Q2 2016 employed 1,804 FTE collectors.  2,294 Q1 2017, Q2 undisclosed)

• Management is opening two new call centers to handle additional debt purchasing

Yet there is a disconnect in results:

• Earnings reports continue to disappoint in 2017, described as poor by all accounts

• The 2017 Americas Core vintage is booked at a 195% purchase price multiple, the lowest starting multiple (lowest estimated collections per purchase price) ever recorded, despite the improved environment and management actions to ramp the business.

Is the company broken? Perhaps.  Historically, amortization rates jump the year following allowance charges, reverting in 1-2 years’ time.  See 2009-2013 Americas Core results for example, with 2015-2016 having similarly high allowance charges.  Here it is critical to remember cash collections allocated to amortization are not taxed in GAAP accounting.  My thesis is management has further amplified the conservative nature of their assumptions due to 1) cover provided by 2015-2016 CFPB-inspired allowance charges and 2) a desire to minimize/delay cash taxes due to single-party control of government and the possibility, however small you may judge it to be, of tax reform.  A temporary side benefit regardless of tax reform would be increased tax-free reinvestment of cash collections into new purchasing during 2017.  I believe management has a history of actively minimizing its tax burden.

Here’s the full portfolio amortization rates, ex-allowances (w/ allowances in parentheses):

FY 2012: 40.9%   (41.6%)

FY 2013: 42.3%   (41.9%)

FY 2014: 41.8%   (41.4%)

FY 2015: 41.9%   (43.8%)

Q1 2016: 43.7%   (46.3%)

Q2 2016: 44.0%   (47.3%)

Q3 2016: 41.9%   (45.5%)

Q4 2016: 44.2%   (62.2%)

FY 2016: 43.5%   (50.1%)

Q1 2017: 48.1%   (48.8%)

Q2 2017: 48.2%   (49.1%)

Note the large jump in Q1 2017 and how it appears “in line” historically if calculated with allowances included, as it is by default.  If the tax thesis is correct, we should see increased amortization charges in US portfolios and not just from Insolvency or European collections where amortization rates are naturally higher.

Quarterly amortization rates and charges, by segment.  There is some seasonality, so compare Q1 to Q1:

Segment: Q1 2016, Q2, Q3, Q4, Q1 2017, Q2 2017

Americas Core: 37.5%, 36.4%, 33.1%, 33.6%, 42.2%, 40.4%

Americas Core: 82.2m, 77.8m, 69.7m, 64.9m, 95.7m, 87.6m

Note the large jump in amortization in 2017 on essentially similar cash collections ($443m 1H 2017 vs $433m 1H 2016, an increase of ~$24m amortization vs ~$10m collections)

Americas Insolvency: 53.5%, 58.0%, 60.6%, 65.6%, 62.3%, 67.8%

Americas Insolvency: 36.8m, 39.3m, 36.6m, 34.8m, 31.0m, 36.0m

Euro Core: 50.4%, 49.8%, 47.9%, 52.2%, 53.0%, 53.3%

Euro Core: 47.5m, 51.2m, 46.0m, 50.8m, 51.9m, 52.8m

Euro Insolvency: 73.9%, 69.8%, 74.6%, 76.3%, 78.5%, 74.3%

Euro Insolvency: 1.5m, 1.9m, 3.5m, 3.8m, 3.9m, 3.9m

Dollar figure-wise, we see the biggest jump in the domestic portfolio, which makes sense if the tax thesis is really in play.  The increased purchasing of Americas Insolvency segment has not yet had time to significantly impact collections and amortization rates of the overall portfolio.

Furthermore, Encore Capital (ECPG), PRAA’s peer and other half of oligopoly in the US, has reported the following amortization rates (ex-allowances) in its US portfolio:

2012: 42.9%

2013: 43.4%

2014: 41.6%

2015: 40.5%

2016: 39.0%  (Q1-Q4, respectively: 43.3%, 41.4%, 36.1%, 35.1%)

2017 Q1: 41.5%

2017 Q2: 40.0%

Critically, we don’t see the large jump in domestic amortization rates for ECPG (2016 39% to 2017 ~40.7%) we do with PRAA (2016 35% to 2017 ~41%), indicating that it is unlikely the purchasing environment alone is responsible for PRAA’s estimate changes.  Either PRAA’s debt significantly worsened at fiscal year-end 2016 after all CFPB-impaired debt had been written off, or management’s forecasts became significantly more conservative relative to ECPG’s.  I tend to hold PRAA management in higher regard, so lean toward the latter.

My expectation is for PRAA domestic amortization rates to begin declining after clarity is reached on tax reform (regardless of outcome) or starting in Q1 2018 as portfolio performance dictates improved estimates.  Results in the meantime are likely to continue to disappoint in the $0.25 EPS/quarter range, but the 1x EV/Net ERC valuation should limit downside.  I expect full ‘normalization’ of amortization rates in the 2019-2020 timeframe, though the trend should be clear well ahead of normalization.  ECPG’s full portfolio Q2 2017 amortization rate was 41.0% vs PRAA’s 48.2%.  A long-term convergence in the 43% range or lower is reasonable outside of “the industry is broken” scenarios and would be in-line to somewhat-higher compared to historical results.

For perspective, my 2019 estimated $3.96 EPS comes with 43.7% amortization rates.  A +/- 500bps change in that amortization rate would raise or lower forecast 2019 EPS by approximately $1.30 on identical cash collections.  The primary questions for investing in PRAA is what your base rate for amortization should be, whether you want to wait to see a trend of declining amortization rates before committing capital, and if the opportunity falls in the 'too-hard' pile.

Data sources referenced in text or pulled from PRAA and ECPG company reports, analyst estimates

Catalyst

Declining amortization rates.  Every dollar allocated to or freed up from amortization charges drops directly to pre-tax earnings, creating optics of tremendous operating leverage and growth of GAAP results.

Specifically, I believe amortization rates are artificially high currently as management is booking purchases at overly conservative assumptions to save on cash taxes paid in 2017 while waiting for clarity on corporate tax reform and/or free cash flow generation to catch up to the improved purchasing environment in the United States.  I expect amortization rates to begin to fall in Q1 2018, or as soon as clarity on tax reform is reached, and to be significantly lower by 2019-2020.  This will create growth in reported GAAP numbers in excess of fundamental cash collection growth.