Citi Could Face A $22 Billion Loss On Put-Back Mortgage Bonds

Now that Bank of America has been hit with a giant lawsuit from mortgage bond investors demanding the bank repurchase loans, investors will scramble to figure out the levels of exposure at every other major bank.

Mark Lennihan

So how large is Citigroup’s exposure? Using some figures from Citi's earning's presentation, and financial analyst Dick Bove's formulas, I've concluded that there may be as much as $58 billion in potential exposures in bubble-era home loans packaged in mortgage backed securities serviced by Citi. Citi may be liable for buying back as much as $35 billion of the loans. After recovering some of that through foreclosure sales, Citi may be looking at a loss of $22 billion.

A few caveats are in order. Citigroup will likely fight the demands that it repurchase the loans. The lawsuits will take years, which is both good and bad for Citi. It's good because it means put-backs are unlikely to quickly punch a hole in Citi's balance sheet. It's bad because it means this could weigh on earnings for years and legal costs will keep mounting.

Most importantly, these numbers are only my best estimates of Citi's future put-back liabilities. I cannot say for certain--no one can--what the ultimate cost of the put backs will be for Citi. The number could be far smaller than our estimate. Or it could be larger.

My starting point is trying to estimate the total size of Citigroup's securitization business during the worst years of the housing bubble. On the earnings call yesterday, CFO John Gerspach discussed a portfolio of loans that Citi services but does not hold. His remarks suggested that Citi is using this portfolio of loans as a way of putting a framework around their total third-party put-back risk.

My methodology is to apply the lessons of my article yesterday, on Dick Bove's analysis of the aggregate cost of bank putback risk, to the loan portfolio discussed below.

During the earnings presentation, Gerspach said the following about a portfolio of loans "serviced but not held" by Citigroup:

"Loans serviced, but not held, which best represents our outstanding loan originations held by third parties. Our total portfolio was $504 Billion, at the end of the third quarter. Of which approximately 3% represents loans securitized to private investors. The reps and warranties are generally issues that arise out of the delinquent portfolio. So, to size that, 8.1 % of the portfolio is currently 30 or more days delinquent. Approximately 22% of the servicing portfolio is covered under indemnification agreements related to acquired portfolios, and another 24% of the portfolio was originated in 2005 or before. So only a third of the portfolio represents unindemnified loans originated in 2006-2008, which have generated the bulk of our claims and repurchase activity."

What does Gerspach's statement tell us about the Serviced but Not Held Loan portfolion held at Citi? (First, let's coin the acronym before someone beats us to it: Let's call this class of loans SBNH Loans.)

Based on Gerspach's remarks, it certainly sounds like Citi is using these SBNH loans--or the riskiest subset of SBNH loans, at least--as an index of, or proxy for, Citi's putback risk.

(Using a portfolio of SBNH loans in this fashion seems a curios notion to me: The servicing agent on a loan is not necessarily the same as the originator (or the securitizer, for that matter) of the loan--who should rightly assume the putback risk, although one might imagine a there significant of overlap on a theoretical Venn diagram. But if Citi wants to use that number as an index of putback exposure, it sounds like a reasonable starting point.)

Based on the 'Servicing Portfolio' called out in the stacked bar graph below, we can see a total of $504 billion in SBNH loans held by Citi Group. Let's break out the 2006-2008 vintages, apart from the less risky vintages and indemnified loans, as spelled out by Gerspach in his remarks.

That leaves us with a total of about $166 billion in SBNH loans--with vintages of 2006, 2007, and 2008. Those SBNH loans break down as follows:

  • 2006 Vintage: $45.0 billion--or 9% of Citi's SBNH Loan portfolio
  • 2007 Vintage: $55.4 billion--or 11% of Citi's SBNH Loan portfolio
  • 2008 Vintage: $65.2 billion--or 13% of Citi's SBNH Loan portfolio

Let's return to the total $166 billion figure for all SNBH loans with vintages of 2006 through 2008.

As spelled out in my article yesterday, Bove assumes a 35% default rate on the riskiest loan vintages, versus a 14% industry wide default rate. He further assumes that 60% of those loans can be put back to the banks. After the banks have had the mortgages put back to them, Bove assumes a 65 percent loss.

So on the $166 billion in total SBNH loans listed above, assuming a default rate of 35%, we would arrive at about $58 billion in defaults. Applying Bove's 60% put back rate, we would get a total dollar value putback to Citigroup on the SBNH loans of around $35 billion.

Finally, using Bove's final plug number of a 65% loss on the put-backs, Citi would be looking at a total loss of about $22.75 billion from the 2006-2008 SBNH loans alone.

That $22.75 billion loss number is an eye-popping 200+ times higher than the $100 million realized loss expressed on Citi's Consumer Mortgage Reps and Warranties slide.

And remember, the SBNH loans may not represent the totality of risk exposure that Citi has to residential mortgage put-back.

What model will best predict the risk exposure banks will face to residential mortgage putbacks? Of course, that's something we'll only know for sure in retrospect--but what seems highly likely is that we can look forward to a lot more of this parsing and analysis in the future.

Calls requesting comment from Citi on Tuesday were not immediately returned.


Citigroup chart


Companies mentioned in this post

Calls requesting comment from Citi on Tuesday were not immediately returned.

Companies mentioned in this post

Bank of America



Questions? Comments? Email us

Follow NetNet on Twitter @

Facebook us @