Calculating Spreads for Valuation, Competition & Opportunity in Unsecured Lending

by | Aug 26, 2018 | Blog | 0 comments

Greetings,

US stocks look set to enter the longest-ever bull market. The current bull market at 3,453 days since the low in March 2009 is now tied with the 1990-2000 bull market, and is on pace to exceed that target.

Meanwhile, in the minutes of the July Fed meeting released this week, the Fed affirmed its stance to gradually raise interest rates. Strong economic growth and decades-low unemployment have kept the Fed on track to raise rates at least once more in 2018. Rising front-end rates have flattened the yield curve. The spread between the 10-yr and 2-yr treasury yields is now down to just 21 bps, and an inverted yield curve has usually been a reliable predictor of recession

In FinTech financing news, Upgrade, Renaud Laplanche’s fintech personal loan platform has raised $62 Mn in a Series C round that was led by CreditEase, valuing Upgrade at $500 Mn. Bungalow, a residential real estate rental platform, has raised $64 Mn in Series A funding led by Khosla Ventures to expand operations.

In this week’s newsletter we continue our series on valuation. This week, we dive into PeerIQ’s methodology to derive the appropriate discount rate for each loan. We also look at the rising competition in unsecured lending, Prosper’s Q2 2018 results and provide a glimpse into our upcoming 3Q2018 Lending Earnings Insight report that we will be releasing next week.

Competition – and Opportunity – in Unsecured Lending

Competition for personal loans is heating up. Mailings soliciting personal loan borrowers exceeded credit card mailings in six of the last seven months, even though the credit card industry is almost 6 times larger. Goldman Sach’s Marcus has been one of the new competitors having originated nearly $4 Bn in personal loans since their launch – approximately a $2 Bn annual run-rate – smaller than Lending Club and others, and likely tempering their ambitions due to late-cycle credit conditions. Marcus recently launched in the UK with personal loans and savings accounts.

As seen in the chart below, growth in personal loan balances has significantly outpaced that other asset classes, averaging near 20% YoY. Outstanding personal loan balances now stand at nearly $125 Bn.

Source: TransUnion

Source: TransUnion

Prosper’s Q2 Results – Originations Up, Net Loss Down

Prosper reported Q2 2018 results that showed growth in originations of 12% YoY to $867 Mn. Net revenue was flat at $32 Mn while net loss dropped significantly from -$41 Mn to -$13 Mn YoY. Prosper also increased its warehouse facility to $200 Mn.

Prosper has also raised interest rates on its loans. As you can see in a screenshot from our Pool Insights page, the median interest rate on Prosper loans increased by 60 bps to 13.59% in Q2 2018 QoQ. Over the same time, the median FICO score for Prosper originations remained steady at 709 and both the 12-month and 2-year LIBOR rates sold off by ~35 bps. This indicates that the implied spread on Prosper loans ceteris paribus increased 25 bps QoQ.

Source: PeerIQ

How to Calculate the Appropriate Discount Rate

In last week’s newsletter we looked at how we generate loan-level cashflows – the numerator of the discounted cashflow / present value equation. This week we will derive the appropriate discount rate for each loan – the denominator.

You can also read our non-technical piece on the top 5 questions to see if your whole-loan valuation makes sense here.

Valuation Framework

The price of a loan is the present value of the loan’s future cashflows (after accounting for losses and prepayments), discounted by an appropriate discount rate:

Once we know the value of each loan in the portfolio, we can then calculate portfolio value by summing up individual loan values. After calculating the cashflows using our machine learning approach, we derive a discount rate to arrive at the price of each loan.

Discounting Framework

Marketplace lending loans, unlike other fixed income securities like Treasuries, have both prepayment and default risks. Investors earn compensation for bearing these risks in the form of a spread above a benchmark curve.

There are four main steps in calculating the discount rate of a loan:

  1. Determining the benchmark interest rate curve to use
  2. Calculating the loan’s spread (above the benchmark) at origination (SATO)
  3. Adjusting the credit spread as the loan seasons
  4. Incorporating market movements (yield curve shifts, spread changes, market transactions, etc.) into the underlying benchmark rate

Determining the Benchmark Interest Rate Curve

A loan’s schedule of cashflows is discounted according to the appropriate discount rate corresponding to the tenor.

For example, a risk-free cash flow expected in 5 months will be discounted using the 5-month zero coupon risk-free rate. Given that these loans have prepay and default risk, we must add a spread to the risk-free benchmark to compensate investor’s for this uncertainty. For every loan, each projected monthly cashflow is discounted to the present day using the sum of a benchmark rate curve and a zero-volatility credit spread.

Typically, swap or treasury rates can be used as benchmark interest rates to derive the discount rate. Here we use swap rates as a benchmark curve to account for lenders’ financing terms which are generally tied to swap rates rather than risk-free Treasuries.

Calculating the Spread at Origination

Investors often buy newly originated loans from platforms at par in an arms-length transaction. For example, an investor funds or buys a $10,000 unsecured personal loan from a lender and pays $10,000 for the loan. This is the first (and often last) observable “traded” price of the loan.

We use this market transaction to determine the credit spread at origination by plugging the transaction amount into our present value calculation and solving for spread given a set of forecasted cashflows. The credit spread at origination (SATO) can be interpreted as the incremental return an investor earns for bearing default and prepayment risk for a newly issued loan over the swap curve.

The SATO is a zero-volatility spread over the benchmark swap curve that equates the present value of the cashflows to the market-observed price. Since the benchmark is considered a risk-free rate, SATO is effectively a credit spread due to the uncertainty of future cash flows.

Exhibit 1: SATO calculation

Source: PeerIQ

Exhibit 2 shows boxplots of sample SATOs for loans originated in December 2015 for one specific originator. The box plots are arranged from low to high credit score cohorts (10 credit score bands) at the time of origination. As expected, borrowers with low credit scores exhibit higher variance in SATO distribution and higher average credit spread, reflecting a higher risk premium.

Exhibit 2: SATO by Credit Score Buckets for December 2015 Loans

Source: PeerIQ

Calibrating a Spread Decay Curve

As loans season, the variance of their expected cashflows decreases. To account for the higher certainty of expected cashflows, we tighten the credit spreads as loans season. As we cannot observe the prices or credit spreads for seasoned loans (because loans are not marketable securities with a liquid aftermarket), we need a methodology to adjust the SATO as the loan seasons. PeerIQ uses the observed term structure of CDX HY and IG indices to adjust the credit spread.

CDX HY and IG indices reflect high-yield and investment-grade corporate credit risk respectively, and act as a proxy for the market’s view on credit risk.

We argue that CDX is an excellent proxy for credit risk in the market because CDX indices are:

  • liquid and widely accepted
  • correlated to the US economy and other macro drivers
  • the only observable credit term structure benchmark available
  • tradeable
  • the most widely used credit hedge for mortgage and asset-backed securities

The CDX index is also imperfect – it captures corporate high-yield risk rather than consumer credit risk. Since there is no consumer credit index, this is the best alternative.

To calibrate the term structure of credit spreads on a loan, we use the SATO on the loan and the CDX HY and IG term structures at the time of origination. As the loan seasons, we adjust the z-spread using SATO as a starting point and the CDX HY and IG term structures as a proxy.

Exhibit 3: Term Structure of Credit Spreads

Source: PeerIQ, Bloomberg

The observable credit spread decay pattern in the corporate bond market is used to calibrate and estimate the unobservable credit spread decay in the consumer loan market. We then use the CDX term structure to create a calibrated credit term structure for each loan with SATO as the starting point.

Exhibit 4: Spread Curve Calculation

Source: PeerIQ

The scaling factor at origination is the ratio between the SATO and the CDX HY spread for a tenor corresponding to the loan’s original maturity (750 bps / 353 bps = 2.12 in the chart above). We then adjust that scaling factor as the loan matures to correspond with the change in spread we see for CDX indices.

Credit Decay Curve Calculation Example

Below we look at a hypothetical loan with a SATO of 750 bps and an original term of 60 months. The loan has now seasoned for 12 months, and we can calculate a new credit spread for the remaining 48 months using the CDX term structure observed at the origination of the loan and its SATO.

We select the appropriate CDX spread by interpolating points between indices and using the IG 5yr as a proxy for the credit spread at maturity.

 

We then interpolate the spread to be used in month 12 for the remaining 48 months using the CDX HY 3YR and 5yr spreads.

The decay factor is the ratio of the interpolated HY spread and the HY spread for the original term of the loan.

We then multiply the SATO (which was calculated for an original term of 60 months) by the decay factor, to get the spread at month 12 for the remaining 48 months.

This is one of the many ways in which credit spreads can be adjusted to take loan seasoning into account for valuation purposes.

 

Next week, we will conclude by applying these valuation methodologies to examine the price profiles of MPL personal loans as they season.

Reach out to learn how PeerIQ can help you with valuations.

 

PeerIQ’s Lending Earnings Insights

Next week we will release our 3Q2018 Lending Earnings Insights report that tracks the quarterly earnings of bulge-bracket banks, credit card issuers and fintech and non-bank lenders. Some of the key themes that we explore this quarter are:

  • We remain in the late stages of the credit cycle with strong economic growth and most lenders seeing low charge-offs.
  • We see divergent credit performance across FinTech asset classes. Subprime and small-business lending is seeing near cycle-low charge-offs but there is a marginal pick-up in charge-offs in the prime sector.
  • Card issuers are increasing loan loss reserves at a higher rate than loan growth, indicating expectations of higher losses going forward.
  • Banks are competing to win the digital consumer, in different ways. Some banks see the mobile app as central to managing customer relationships while others see it more as a cost-saving mechanism.
  • Lenders are looking to manage the impact of rising rates and have managed to eke out NIM growth in this rising rate environment.

Conferences:

Industry Update:

Lighter Fare: