J.P. 摩根：消费者金融2018年展望.pdf

jpmorganmarketsNorth America Equity Research07 December 2017Equity Ratings and Price TargetsMkt Cap Rating Price TargetCompany Ticker ($ mn) Price ($) Cur Prev Cur PrevCapital One COF US 45,470.63 93.87 OW n/c 100.00 n/cAlly Financial ALLY US 12,377.47 27.89 OW n/c 32.00 29.00Sallie Mae SLM US 4,962.53 11.49 OW n/c 14.00 n/cPennantPark Investment PNNT US 518.74 7.30 OW n/c 8.50 n/cTwo Harbors TWO US 2,815.73 16.14 OW n/c 17.00 n/cAmerican Express AXP US 86,523.01 98.21 OW n/c 105.00 102.00Source: Company data, Bloomberg, J.P. Morgan estimates. n/c = no change. All prices as of 06 Dec 17.Corrected Note (See page 35 for details)Specialty & Consumer Finance2018 Outlook: Extra InningsSpecialty & Consumer FinanceRichard Shane AC(1-415) 315-6701richard.b.shanejpmorganBloomberg JPMA SHANE Melissa Wedel, CFA(1-415) 315-6763melissa.wedeljpmorganJ.P. Morgan Securities LLCSee page 35 for analyst certification and important disclosures.J.P. Morgan does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of this report. Investors should consider this report as only a single factor in making their investment decision.One of Wall Streets favorite clichd questions seems to be “what inning are we in?” In keeping with that metaphor, we believe we have entered extra innings of an epic game. Looking down the lineup, it is possible to imagine the game continuing. Low unemployment, nascent wage growth, steady home price appreciation, a benign regulatory environment, and tax reform create a potentially compelling backdrop (i.e., equivalent of deep bullpens). But we never want to forget that once a game is in extra innings, it is only a pitch from being over. A rebound in volatility, the Fed unwind, inflated financial asset prices, negative sovereign yields, credit normalization and political chaos can be viewed as veteran sluggers lining the benches. NB: Not only big bats can end games; a hanging knuckleball in the 11th inning to a .254 hitter may end the season in tears.With that in mind, we believe investors should begin to de-risk during 2018. We dont advise going home before the game is over, but it is probably not a great time to order another round of dogs, if one wants to beat the traffic out of the stadium when the game ends. Credit Cards Growth math and timing will likely drive array of results in 2018. Please refer to “Deep Dive Growth Math” beginning on p.3 for detailed analysis of seasoning curves on credit card profitability. Increase AXP PT to $105 on 14.5x multiple of 2019e EPS (from $102 on 14.0x multiple), driven by spend strength.Increase SYF PT to $39 (from $37) on 2019 EPS accretion from anticipated acquisition of PayPal portfolio.Overall top pickCOF Auto Finance Auto credit appears to be stabilizing. Two years of tighter underwriting and reduced supply of used cars from hurricanes (combined with labor and wage stability) suggest auto NCOs are approaching cyclical peak consistent with our predictions of a moderate cycle. Increase ALLY PT to $32 on 1.0x multiple of 2019e adj. TBV (from $29 on 0.9x multiple), driven by stable credit outlook. Auto Finance pick ALLY Student Lenders Hope of student lending reform has receded to expectation of status quo. Please refer to “Deep Dive Cumulative Loss Metrics” on p.10. As a result, student lenders offer best relative value within our coverage.Student Lending top pick SLMCompleted 07 Dec 2017 01:06 AM ESTDisseminated 07 Dec 2017 01:07 AM EST2North America Equity Research07 December 2017Richard Shane(1-415) 315-6701richard.b.shanejpmorgan Commercial Finance Competition has been strong for new issuance in 2017 and we expect more of the same into 2018. Spreads are narrow and we assume spreads will level off in 2018. While this would likely lead to NAV stability, we believe new investment yields and interest income may remain pressured. Commercial Finance top pick PNNT Mortgage REITs Stable housing credit and low volatility/benign rate outlook provide a favorable backdrop. Performance of credit and rates strategy may converge in 2018 as end of Treasury re-investment offers incremental opportunities for agency investors. Mortgage REIT top pick TWO3North America Equity Research07 December 2017Richard Shane(1-415) 315-6701richard.b.shanejpmorganCredit CardsDeep Dive Growth MathOver time, outcomes in the consumer finance sector are ultimately determined by credit selection. In our view, the best and most durable franchises are able to demonstrate consistent risk-adjusted margins. We also believe credit performance is efficient and that NCOs are highly correlated within verticals and are tied to cyclical factors (i.e., we tend to think of lenders in the context of sensitivity to the cycle, or “credit beta”). In the intermediate term, factors like asset growth, investment, reserve build and loss migration can distort results. For example, we estimate that a “normal” vintage of card loans with a lifetime ROA of 2.5% has a GAAP ROA that varies between 0% and 4% over the course the first 5 years. Consequently, growth can create short-term earnings distortions, particularly if it contributes to higher concentrations of pools of loans at their trough ROAs. NB: This analysis holds lifetime ROAs for the vintages constant.The table below is taken from Discovers ABS Static Pool Reports. While we recognize there are limitations to the data, it is one of the better illustrations of credit card loan seasoning. Based upon these loss curves, we estimate card issuers may incur as much as 2/3rds of their static pool losses in the first three years of each vintage. We draw upon the shape of these seasoning curves in the analysis below.Table 1: Discover Card Master Trust, Static Pool NCO Rate Source: Company reports and J.P. Morgan estimates.To better understand the life cycle of a vintage, the analysis below illustrates the ROA of a hypothetical vintage over a 5-year time horizon. Nine Months EndedSept. 30Origination Year 2017 2016 2015 2014 20132017 0.00%2016 5.52% 1.58%2015 6.11% 4.61% 0.00%2014 5.91% 5.70% 0.00% 0.00%2013 5.25% 5.44% 0.00% 0.00% 0.00%2012 and prior 1.91% 1.64% 1.64% 1.76% 1.89%Total 2.10% 1.71% 1.64% 1.76% 1.89%Per Discover filing: Additional accounts were last designated for the mastertrust as of July 1, 2016, with the previous such designation as ofOctober 1, 2008. Because the additional accounts designated on July 1, 2016 did not include any charged-off accounts, onlyaccounts that have become charged-off accounts since deisgnation can generate recoveries. Therefore, net charge-off ratesshow n for the accounts in origination years 2012 through 2016 may not reflect the benefit of recovery rates that are comparable toaccounts that have been designated prior to July 1, 2016Twelve Months Ended Dec. 31, 4North America Equity Research07 December 2017Richard Shane(1-415) 315-6701richard.b.shanejpmorganTable 2: Static Pool Results Over 5-Year HorizonSource: J.P. Morgan estimates.In the first part of this analysis, we track the ROA of a single vintage over its first 5 years. This methodology allows us to build a simple income statement and balance sheet for the pool (see Table 2, above). In this example, we assume the following: Initial pool of loans: $100M; Loss curve ramps to a peak loss rate of 6.25% in year 3 (and declines to 3.5% in year 5); Net revenue margin (interest income + fees interest expense) = 15%/year; “Pool” efficiency ratio starts at 60% and improves to 50% during year 3 as marketing/rewards expense declines and remaining portfolio positively selected for better credit (e.g., lower servicing costs); Reserves levels are consistently held at next 5 quarters of forward losses (a relatively conservative reserve policy in context of industry norms).The primary factor explaining the ROA variance during the first 5 year is the seasoning of credit performance. We believe that the impact of loss seasoning is generally underappreciated. We estimate that over the first five years, a vintage with an annualized 3.9% NCO rate will report peak losses of 6.25% near month 36 before losses recede to 3.5% in year 5 (see Figure 1, blue line). GAAP results are then further complicated by the forward-looking nature of reserve methodology, which requires lenders to typically hold reserves of 12 to 15 months of forward losses. In this analysis, provision expense peaks near month 24 (see Figure 1, gray line).0 1 2 3 4 5Balance SheetGross Loans 100.0$ 103.0$ 103.5$ 101.5$ 99.9$ 99.2$ = Portfolio grows 1% quarter, net of charge-offsReserve (1.0) (4.4) (6.9) (6.3) (5.0) (3.9) = Reserve equals 15 months projected lossesNet Loans 99.0$ 98.7$ 96.6$ 95.2$ 94.8$ 95.3$ Coverage ratio 1.0% 4.2% 6.6% 6.2% 5.0% 3.9% = Coverage ratio ramps ahead of NCOsIncome StatementNet Revenue 15.1$ 15.5$ 15.5$ 15.2$ 14.9$ 14.9$ Provision (4.1) (5.9) (5.8) (4.7) (3.6) (3.5) Op. Ex (9.1) (8.5) (8.5) (7.6) (7.5) (7.5) Pre-tax income 1.9$ 1.1$ 1.1$ 2.8$ 3.9$ 4.0$ Net Revenue Margin 15.0% 15.0% 15.0% 15.0% 15.0% 15.0% = Constant over lifeEfficiency Ratio 60.0% 55.0% 55.0% 50.0% 50.0% 50.0% = Marketing spend & servicing intensity declineROA (year) 1.9% 1.0% 1.1% 2.8% 3.9% 4.0%Year5North America Equity Research07 December 2017Richard Shane(1-415) 315-6701richard.b.shanejpmorganFigure 1: Provision Expense versus NCO Rate (single vintage, 5-year time series) Source: J.P. Morgan estimates.This timing differential creates a significant distortion between the timing of credit expenses and net revenues. Based upon this analysis, a card issuer will recognize 57% of expected provision expense in the first 2 years of a vintage, versus only 39% of revenues. This trend begins to reverse in year 3.Figure 2: Cumulative Provision Expenses vs Cumulative Net Revenues (as % of 5 years)Source: J.P. Morgan estimates.The earnings impact of these differing factors is illustrated below. Figure 3 is a time series, tracking credit seasoning (front loaded in first three years) with profitability (back-end loaded in years 4 and 5). We specifically note that a vintage with an average ROA of 2.5% over 5 years may deliver a 4%+ ROA after year 4.0.0%1.0%2.0%3.0%4.0%5.0%6.0%7.0%0 1 2 3 4 5NCO Provision ExpenseProvision expensepeaks at 24 months NCO rate peaks at 36 months57%43%39%61%0%10%20%30%40%50%60%70%Months 1 to 24 Months 25 to 60Provision Expense Net Revenues6North America Equity Research07 December 2017Richard Shane(1-415) 315-6701richard.b.shanejpmorganFigure 3: NCO and ROA (static pool over 5-year horizon)Source: J.P. Morgan estimates.Figure 4: Y/Y Credit Card Growth (quarterly, past 3 years)Source: Company reports and J.P. Morgan estimates.Early Retail Sales Bode Well for 4Q17Retail sales are off to a solid start in the holiday shopping season. The National Retail Federation expects holiday spending to total $680B, which would reflect a 4% Y/Y increase. Online sales continue to take share from brick and mortar retail stores. Adobe Analytics reports that online purchases in November through Cyber Monday totaled $50B, +17% Y/Y, and that online sales exceeded $1B every day in November. Meanwhile, research firm ShopperTrak noted that store visits on Black Friday were down 1.6% Y/Y. 0.0%1.0%2.0%3.0%4.0%5.0%6.0%7.0%0 1 2 3 4 5NCO ROA (pre-tax) NCO (Avg.) ROA (Avg.)0.0%2.0%4.0%6.0%8.0%10.0%12.0%14.0%16.0%AXP (ex-Costco) COF (ex-Cabellas) DFS SYFCOFs growth has slowedpast 6 quartersDFSs growth has rampedpast 4 quarters7North America Equity Research07 December 2017Richard Shane(1-415) 315-6701richard.b.shanejpmorganFigure 5: Online Sales (2017)$ in billionsSource: Adobe Analytics.Figure 6: Number of Transactions by Day (2017)Source: Adobe Analytics.Figure 7: Black Friday Market Share of TransactionsSource: Adobe Analytics.Figure 8: Cyber Monday Market Share of TransactionsSource: Adobe Analytics.In addition to Y/Y spending growth during the key holiday season, early indicators suggest shoppers are being offered fewer discounts than in prior years. The average Black Friday advertised discount across 17 categories was 45% this year compared with 48% last year, according to price-tracking firm Market Track LLC. Only three of eight major retailers offered deeper discounts than they did a year ago. Meanwhile, macro factors serve as a tailwind, as consumers remain healthy. Unemployment is low and debt levels are manageable. Consumer confidence climbed to 17-year highs consumer confidence index rose to 129.5 in November,up from 126.2 a month earlier.Top Pick: Capital One (COF)Capital One has entered the “harvest” window from timely investments in 2015 and 2016. We believe that in 2018, the 2015/16 vintages transition from being dilutive to ROA to modestly accretive (see Figure 3 above for our generic time series analysis of a generic static pool ROA). We note that COFs US Card portfolio grew more than 2x the industry rate in 2015 and 2016. Going forward, the addition of $5.7B in high-quality assets from Cabelas should enhance efficiency and moderate the NCO ramp (10-15bps benefit). COF has added $3.0B to US Card reserves, an increase that is roughly 50% above that of the closest competitor and 3x the industry average. Coverage ratios (as % of net receivables and % delinquencies) are at/near top of industry. Our 2018 and 2019 estimates are above consensus by $0.05 and $0.17, respectively, and are positively levered for upside from tax reform and wage growth.$2.9$5.0$6.6Thanksgiving Black Friday Cyber Monday5.6 7.18.62.9 3.23.112.5 12.9 14.5051015Thanksgiving Black Friday Cyber MondayAmazon Walmart Target Best Buy Kohls Others TotalAmazon, 54.9%Walmart, 8.8%Best Buy, 5.0%Kohls, 3.4%Target, 3.3%Others, 25%Amazon, 59.6%Walmart, 8.6%Target, 4.6%Best Buy, 3.1%Kohls, 2.7%Others, 21%8North America Equity Research07 December 2017Richard Shane(1-415) 315-6701richard.b.shanejpmorganAuto FinanceResidual Values and Auto Sales Rebound in Wake of HurricanesThe 2017 hurricane season, among the top 5 on record, has resulted in used car values rebounding after 1-2 years of declines and higher retail sales. The Manheim US Used Vehicle Value Index hit an all-time high of 136.3 in October, driven by increased demand and reduced supply for vehicle replacement in hurricane-affectedareas. Given the nature of how the Manheim index is calculated, the value is expected to grow over time, but the acceleration post-hurricane season is significant. The JD Powe