At current near-record levels of consumer mortgage rates, and with MBS prices grinding higher, a natural question is whether rates are poised to push downward to new lows. In turn, a critical issue is whether the market for 30-year conventional 3.0s has become liquid enough to absorb a substantial increase in issuance of loans with rates of 3.75% and lower. I’m dubious; as I’ll show, the lack of issuance and thin float in 30-year Fannie and Gold 3s suggests to me that a sustained burst of issuance would crush its price and cause real difficulties for originators attempting to hedge their pipelines.

A brief review of pooling economics will be helpful at this point. The key decision by originators for all loans is the coupon into which they can be originated (or “slotted”). The decision is fairly straightforward. After accounting for the 25 basis points in required servicing that must be held, the execution question is whether to pool into the closest possible coupon or the coupon 50 basis points in rate lower. A key issue is how the guaranty fee is paid to the GSE in question. The question is whether optimal execution results from directing part of a loan’s periodic interest payments toward paying the g-fee over the life of the loan or, alternatively, by paying the guaranty fee in the form of a single payment to the GSE (i.e., “buying down” the g-fee).

As an example, let’s take a conventional loan with a 4.25% rate to the borrower. After accounting for the 25 basis points in required servicing that needs to be held (or sold), the originator has two main securitization options:

• Pay the g-fee over time out of the loan’s interest, hold/sell some excess servicing, and pool into a 3.5% pool; or
• Buy down the g-fee (by making a single payment to the agency) and pool into a 4% pool.

The fee to be paid to the GSE for buying down the g-fee is in turn a function of two factors: the amount of the guaranty fee and the price (quoted as a multiple) that the GSE will pay for that cash flow. Combined with the notoriously expensive multiples quoted by the GSEs to buy down g-fees, the increase in g-fees that went into effect in April makes it uneconomically expensive for originators to buy down g-fees. This in turn means that originators now have a strong incentive to pool down (i.e., into lower coupons) when judging best-execution for their conventional loans.

This brings us to a discussion of Fannie and Gold 3s. Originators would love to see robust and liquid trading in conventional 3.0s so that they can clearly slot their 3.75% and lower loans into the coupon. However, I’m not optimistic that the coupon is liquid enough to support even a modest burst of issuance. Liquidity always has a chicken-and-egg element; it’s difficult to issue into an illiquid security, but liquidity can’t improve without a significant amount of the security available to trade (i.e., a sufficient “float”). In this context, trading in conventional 3s remains problematic. Over the last six months, Fannie and Freddie have collectively issued 158.2 billion in 30-year 3.5s; over the same period, total issuance of 3s was 4.1 billion. The available float of the coupons paints a similar picture. While roughly 194.7 billion in 30-year 3.5s are available to trade (and relatively few have been locked up in agency Remics), total float in 3s in March was 4.2 billion. 

(Read more on this relative lack of liquidity: Time To Start Looking At 3.0 Coupons?)

Of course, all securities must go through this “liquification” process. Two years ago, for example, total float in conventional 3.5s was only 1.4 billion; it only exceeded 100 billion in January of this year. What is different is the lack of natural buyers of 30-year 3.0s. The combination of their low coupon rates and long durations make them unappealing to banks and depositories, and they are very difficult to structure into Remics. (It’s interesting to contrast them with 15-year 2.5s. The total float for conventional 2.5s is a more substantial 12 billion; moreover, there are natural buyers of the shorter-duration 15-year security, even though its coupon rate is quite low.)

As a result of these considerations, I don’t expect the liquidity in 30-year 3.0s to improve enough in the near future to make them a legitimate securitization coupon. A pop in issuance would likely crush the coupon’s price, and cause its day-to-day price performance to return to its erratic pattern seen earlier this year. This in turn will make it difficult for consumer mortgage rates to make a decisive push to new lows.