Equify

Lending Club's Underwriting and Performance

2012-08-15T15:56:00.001-07:00

I've taken a good look at Lending Club - it's an incredible business. They're forced by regulators to file as a public company - and as such, they actually publish extensive details about their underwriting process. The following section is based on information in the Lending Club Prospectus dated July 31, 2012.

According to Lending Club’s prospectus, the base interest rates is set by taking the average of the interest rate for unsecured credit and the average interest rate for a certificate of deposit as published by the Federal Reserve. A committee then adjusts this rate to set a ‘base rate.’ Lending Club assigns borrowers to one of thirty-five different ‘sub-grades.’ The default rate is estimated to be equally divided between the assumed default rate of A1 and G5, based on historical performance.

In the most recent prospectus (July 31, 2012), the interest rates charged by sub-grade are as follows:

Underwriting Criteria:

In the Lending Club model, all loans are priced by adjusting the base rate for the following factors:

· FICO Score

· Loan amount

· Open accounts

· Credit inquiries

· Utilization of credit

· Length of credit history

· Loan term

FICO SCORE

Applications are first assigned a sub-grade based on FICO Score. Because Lending club does not accept applications from borrowers with FICO scores less than 660, the credit grades assigned only span from A1-C5. Lower grades are only assigned as a result of a downward adjustment based on factors applied later in subsequent steps of the underwriting process.

LOAN AMOUNT

Second, the sub-grade is modified based on the ratio of the requested loan amount to the Lending Club pre-determined “guidance limit.”

Sub grades are then adjusted based on the following guidance limits:

OPEN ACCOUNTS

Next, the sub-grade is modified according to the borrower’s open accounts (as reported by a consumer reporting agency) and provided the Borrower member has opened at least three accounts, are as follows:

INQUIRIES

Next, the sub-grade is modified based on the borrower’s number of credit inquiries in the last six months (as reported by a consumer reporting agency). Sub-grade modifications based on the number of credit inquiries in the last six months (as reported by a consumer reporting agency) are as follows:

If a borrower has a credit score of 740 or less, the maximum number of credit inquiries allowed is 3, with the same modifier effect. In addition, for 0 inquiries there is a +1, +2, or +3 modifier for grades A, B or C, respectively.

CREDIT LINE UTILIZATION

Fifth, the sub-grade is modified based on the borrower’s utilization of his or her credit limit. Sub-grade modifications based on utilization of credit limit are as follows:

LENGTH OF CREDIT HISTORY

Sixth, the sub-grade is modified based on length of the borrower member’s credit history. Sub-grade modifications based on length of credit history are as follows:

LOAN TYPE/LENGTH

Seventh, the sub-grade is modified based on the term of the loan as follows:

By adding the modifiers to the initial sub-grade, we arrive at the final sub-grade of the requested loan based on the initial credit criteria.

Expected Returns

Page 42-45 of the Lending Club Prospectus describe their methodology for determining estimated default rates.

Based on the Lending Club Prospectus and underwriting model, defaults are scheduled to increase uniformly by sub-grade. The default rates by sub-grade, as represented in the prospectus, are shown below:

Figure 1: Lending Club Interest Rates and Estimated Defaults

However, our analysis of the actual loan performance shows a much messier picture. In the loan data available at https://www.lendingclub.com/info/download-data.action, loans are assigned one of the following “status”.

Charged Off

Current

Default

Fully Paid

In Grace Period

In Review

Issued

Late (16-30 days)

Late (31-120 days)

Performing Payment Plan

For the purpose of our analysis, we examined loans during the most recent 12 month period ending July 1, 2012. This removed all loans with the status of “issued” where no payment performance is available.

3 and 5 year notes	LoanCount	LoanCount%	Loan Amount(000s)	LoanAmount%
Charged Off	239	0.77%	$2,707	0.68%
Current	28780	92.79%	$370,695	93.21%
Default	19	0.06%	$279	0.07%
Fully Paid	1394	4.49%	$16,486	4.15%
In Grace Period	136	0.44%	$1,700	0.43%
Late (16-30 days)	91	0.29%	$1,201	0.30%
Late (31-120 days)	319	1.03%	$4,103	1.03%
Payment Plan	38	0.12%	$549	0.14%
Grand Total	31016	100.00%	$397,719	100.00%

During the same time period, the distribution of loans and default rates at a % of amount funded are represented in the graph below. Please note that the smooth line depicted in the underwriting criteria and expected default rate is not reflected in the actual data and that an anomaly occurs with borrowers with a ratings of G2 to G4.

Because there is a material difference between the underwriting model’s predicted performance by sub-grade and the actual performance by sub-grade, I've shown Lending Club member loan performance based on actual loans rather than the estimated default rates offered in the prospectus. The table below shows default rates and loans more than 15 days late as a total of funded loans.

In the table below, we use the following assumptions to estimate returns:

· All loans more than 15 days late are expected to default, with no principle repayment. Loans not charged off or in default that are more than 15 days late are represented as late.
· All loans less than 15 days late, current, or fully paid are considered ‘performing’
· All loans will be repaid over 36 months using simple interest
· Loans will either be fully repaid over 36 months or will default

Closing Thoughts:

In the selected time period, with some pretty liberal assumptions, Lending Club has generated extraordinary returns. My question is - how do we swap the good loans for bad? Among the hundreds of thousands that are rejected by lending club each year...how do we identify the creditworthy borrowers among them, and help them get the capital they need to get out of the 'debt trap'.

Endorsements

2012-08-15T15:38:00.002-07:00

Using endorsements online is nothing new. Amazon.com uses customer reviews f to help shoppers make an informed buying decision. Ebay sellers readily collect “feedback”...and some now have over 100,000 positive endorsements. Professionals and job seekers alike collect endorsements on LinkedIn in the form of “recommendations” to elevate their professional credibility.

Fundamentally, an endorsement takes information from people who have unique, first-hand knowledge, and presents that knowledge to a marketplace to facilitate better decision making. When making a purchase, a single endorsement typically isn’t given much weight; but many endorsements are often considered trustworthy, reliable sources of information.

There’s more going on when an endorsement is included in a transaction. When a seller in a marketplace opts into a system with mechanisms to capture endorsements, a social contract emerges that provides pressure to meet or exceed expectations. Amazon and Ebay sellers go to great lengths to protect their seller ratings. Borrowers are expected to behave in a similar manner.

Endorsements have been used in peer-to-peer lending and the results, though not scientific, are very compelling. In 2008, netbanker.com published information indicating that borrowers in the Prosper marketplace with one endorsement[1] were 35% less likely to default as compared to borrowers with similar loans and no endorsements. Further, borrowers with more than one endorsement were 50% less likely to default than borrowers with similar loans.

We realize an endorsement alone doesn’t mean much…but a credible endorsement could be extraordinarily meaningful when evaluating a borrower’s credit worthiness.

[1] Endorsers are individuals who offer positive information about a borrower and bid on a borrower’s loan in the Prosper Marketplace.

http://www.netbanker.com/2008/02/prosper_helps_borrowers_tap_the_value_of_their_social_capital.html

The 5 C's of Credit

2012-08-15T15:37:00.001-07:00

There used to be 5 C’s of credit: capacity, collateral, conditions, capital, and character. Character is hard to define and even harder to quantify – however that doesn’t make it any less meaningful when evaluating a borrower’s default risk.

Over the past 60+ years, since the advent of the FICO Score, the credit score has become the predominant method for evaluating default risk. As we all know, the five dimensions of the FICO score (payment history, amounts owed, length of credit history, new credit, types of credit) combine to be highly correlated to a borrower’s probability of defaulting on a loan. Many have successfully developed supplemental scorecards and models to try to outperform the FICO model, but character is still conspicuously absent from most loan applications.

Borrowing and Lending in the US is broken

2012-08-15T15:35:00.004-07:00

The process of borrowing and lending seems like it should be simple. Those with money to lend want to earn a financial return with reduced risk. Borrowers want access to capital at a reasonable cost. But it’s not that easy. Borrowing and lending in the U.S. has become highly inefficient:

· Credit risk assessment is backward, not forward looking. Traditional lending systems in the U.S. only look at a borrower's history when assessing credit risk. This limited view does not adequately merit a borrower’s character. Individuals with high social equity and insight that could positively influence borrowers’ access to capital are unable to activate the value of their social equity for the benefit of their community.

· Consumer credit models have not adjusted for recent macro-economic shocks. Recent years have seen many individuals who otherwise have an excellent credit history and stable employment, but were forced into a short-sale or foreclosure during the housing bust. These individuals may refrain from borrowing; or if they do borrow, must pay higher interest rates. Lenders have no ability to identify these types of creditworthy borrowers among those adversely affected by the economic downturn.

· Conventional loans and credit are not available to all – either because prospective borrowers don’t qualify by traditional measures, or because the costs are prohibitive. Those lacking a U.S. credit history, such as recent immigrants in highly skilled professions, often have difficulty obtaining credit regardless of their prior country credit history and current earnings. Military personnel who move frequently may be denied economic opportunities associated with reasonably-priced capital, turning instead to payday lenders charging up to 400% APY.

History is a battle between debtors and creditors.

2012-08-15T15:34:00.001-07:00

History is a battle between debtors and creditors. Every so often, these two clash, there is a crisis and the whole system is remade.

Philip Coggan, columnist for The Economist

2012-08-15T12:46:00.003-07:00

What am I looking at?

July 11-July 12 Prosper Loans.

Group by % interest rate.

Current/Paid are all loans have made all payments on time.

Bad are loans that are at least 15 days late and may have defaulted or been written off.

What am I looking at?

July 11-July 12 Prosper Loans.

Group by % interest rate.

Current/Paid are all loans have made all payments on time.

Bad are loans that are at least 15 days late and may have defaulted or been written off.

What happened to character?

2012-08-15T12:37:00.002-07:00

Before there were FICO scores and credit tranches, character was used to evaluate credit worthiness. Character has been conspicuously absent from the lending process since we started to trust numbers more than we trust the people. We think it’s time to bring character back to consumer credit.