One of the popular reasons many blame banks for creating so many bad loans during the crisis was that they could merely sell them to someone else. This criticism of the originate-and-distribute model sparked Congress to create a new rule to force banks to keep some skin in the game. By eating some of their own cooking, banks might pay more attention to credit quality, rather than just selling ugly loans to naïve investors. This week, regulators unveiled a preliminary plan for risk retention to the industry for comment. Will it solve the problem or create new ones?

The plan contains three main menu options for what part of their cooking a bank must retain. All three in some way result in the bank holding 5% of the loans they originate, and that risk cannot be hedged. In order to understand the three options, you must understand securitization. So let's start with a very brief tutorial for those who are not familiar with the practice.

The Basics of Securitization

Securitization is one way in which banks can obtain funding for loans. Basically, they take bunch of loans and create a giant pool. Then they sell bonds that pay investors, in some way, based on the payments of interest and principal borrowers pay each month within that pool of loans. In the example that will be used going forward, imagine a $1 billion pool of mortgages. As those borrowers make monthly payments the bank takes that money and pays investors of the mortgage-backed bonds it had sold. In this case, the structure of those bonds is as follows:

  • Class A-1: $250 million
  • Class A-2: $270 million
  • Class A-3: $220 million
  • Class A-4: $200 million
  • Equity (first-loss) Piece: $60 million

The deal will pay principal sequentially, while interest is paid to all bondholders each month. In other words, as principal comes in, it pays the A-1 bonds first, until they're all paid off. Then A-2 bondholders are paid, and so on. As a result, if losses are less than 6%, they won't hit any of the Class A bondholders.

Option 1: Vertical Slice

The first option would require the bank to hold 5% of each piece of the transaction. In this case, there are only two major classes: A and equity. So it would have to hold $47 million of the class A bonds and $3 million of the equity piece. In other words, it could sell the other $893 million class A bonds and $57 million of the equity piece.

Option 2: Horizontal Slice

The next option allows a bank to simply hold 5% of first loss portion. Under this option, the bank would hold $0 of the Class A and $50 million of the equity. It could sell the entire $940 million in class A bonds and $10 million of the equity. If there was less than 5% equity, then it would hold all equity and a portion of the Class A bonds.

Option 3: L-shaped Slice

This is an odd one, and it's a little hard to explain. But essentially, it merges the two above. Here, the bank must hold 2.5% of each of the non-equity bonds and a portion of equity that provides it with a total of 5% of the entire deal. In this case, that translates into $23.5 million in Class A bonds and $26.5 million in equity. It can sell the other $916.5 million in Class A bonds and $33.5 million of equity.

What Will Banks Choose?

One thing is clear: any bank that can sell the equity piece for a reasonable price would probably be crazy not to go with the vertical slice option. Those Class A bonds would often be rated AAA by the rating agencies. We know the raters don't always get it right, but assuming that they do better going forward, then these bonds have very, very little risk. After all, the deal would have to sustain principal losses of at least 6% in order for Class A bondholders to see a loss.

Meanwhile, the horizontal slice option leaves banks with the most risk. Under this retention method, they'll probably sustain a large portion of the losses. It appears that the new rule would allow the banks to sell their equity retained in proportion to how the deal pays down over time. But in practice it might be difficult for banks to sell the equity long after the deal has closed.

For example, again sticking with the scenario above, let's say that the deal ultimately incurs 3% losses. Under the vertical slice method, the bank would incur a $1.5 million loss. In the horizontal slice scenario, however, if a bank could not sell the equity piece in the years after issuance, it would incur a $25 million loss. If it sold half its retained equity by the time losses hit, then it would still incur a $12.5 million loss.

The problem here is that these scenarios don't really require banks to face the same magnitude of risk. A vertical slice will provide banks with far, far less risk than a horizontal slice. The thinking here is probably keeping in mind industry practices. Some lenders normally keep the entire equity piece, because it is not always easy to sell. In those cases, regulators probably want to give them some leeway in not having to also buy bonds from other classes if they already have a sufficient amount of the equity retained. This might even be the motivation for the L-shaped slice, though it doesn't fully accomplish the desired end, as banks would still have far more risk exposure under this method than the vertical slice and would also be forced to purchase some senior class bonds.

But a more reasonable method for the horizontal slice approach might be to require banks to hold a proportionally smaller piece of equity than 5%, since it bears the first loss. This would be particularly relevant in cases where the equity makes up less than 5% of the deal's size. If in the above scenario, for example, if the equity was only 4% and the Class A bonds were 96%, then a 20% loss would result with the bank losing the entire $40 million equity piece, plus another $1.66 million on their A-class holdings. In a vertical slice scenario, however, the bank would only incur a $10 million loss.

Unfortunately, it won't be easy to create rules that more fairly distribute risk that banks face. There's no way to be certain of how much risk each segment of a securitization transaction faces, so you would either have to rely on rating agency grades or banks' internal risk models to provide additional leeway for banks choosing the horizontal slice method. As a result, through this new approach we might see banks having even more motivation to sell as much of a deal's equity pieces as possible in order to utilize vertical slice retention, since it provides them with the least risk. This would actually result in the market taking on more risk.

We want to hear what you think about this article. Submit a letter to the editor or write to