Many investors have expressed concerns about the changing supply/demand dynamics in the MBS market, in light of expected heavy refinancing activity. We will analyze the changing dynamics in some detail, arguing that MBS demand should rise more rapidly than net supply.
A steep curve and a recessionary environment tend to boost bank/thrift MBS demand while also leading to lower net fixed-rate production. This is clearly a positive dynamic for mortgages. Though heavy originations may lead to intra-day volatility in current coupon spreads, we expect that the trend should be for tightening mortgage spreads versus both Treasuries and swaps.
The current environment is very different from 1998, where mortgage spreads widened sharply. This is a credit/Fed led rally in Treasuries, not a flight-to-quality, liquidity crisis (as in 1998). In today's environment, mortgages tend to thrive, benefiting from low funding, and improved convexity cost due to the steep curve.
15-year supply expectations
One concern for the 15-year market has been an anticipated avalanche of supply. However, as we have noted previously, 15-year production (as a percentage of 30-years) has recently been lower than in previous major refinancing waves (1998 and 1993).
We have attributed this to a decline in the outstanding amount of seasoned 30-year premiums, which appear to be the primary source of new 15-years in a refinancing wave. In Exhibit 1 we display the amount of outstanding seasoned 30-year premiums (WALAs longer than 60 months). This implies a declining percent of 15-year refinancing originations relative to the previous refinancing waves.
We recommend a very heavy overweight to 15-years over Treasuries, swaps, agencies or 30-years, favoring 6s.
Bank demand, GSE portfolio growth to absorb supply
In the midst of a refinancing wave, with the Federal Reserve adding cash and liquidity, the supply/demand dynamics of the mortgage market are changing. We have been harping that banks are likely to grow their market share relative to the GSEs.
Historically, banks have increased their MBS holdings during recessionary environments, as loan growth decreases, credit concerns mount and funding becomes more attractive. In Exhibit 2, we display the respective estimated mortgage market shares of the GSEs and U.S. commercial banks and thrifts. We can see that over the last year banks have sustained their MBS market share for the first time in seven years. It is important to observe that though bank market share of MBS has decreased since 1993, the MBS holdings have actually increased from $500 to $700 billion.
The total mortgage portfolio for banks and thrifts is around $2.3 trillion, with $750 billion in securities. The majority of the holdings are in raw loans. The percentage held in securities (relative to raw loans) has been rising recently.
We expect that trend to continue as the economy slows. It has already risen by 1% in the first half of the year. This appears to have primarily occurred through security purchases rather than through securitization of loans.
If we were to estimate a back-of-the-envelope number for bank MBS demand, a first stab might be to assume that the security percentage should rise to around 35%-36% (as it was in 1993). That would require the purchase of about $130 billion in MBS - the current percent of portfolio in securities is around 32%. If 1998 is any gauge, this could occur in the span of a year. It would imply about $11 billion a month in bank purchases.
Current net purchases have been around $7 billion/month in Q2 2001. We estimate that net fixed-rate MBS production should drop to a little over $10 billion a month by the end of 2002, from the current level of near $30 billion/month (see discussion below). These estimates would imply that banks could absorb all of the net supply, without assuming any GSE portfolio growth. This analysis further leads us to have a very positive long-term outlook on mortgages.
Servicing costs rise
Though we anticipate that heavy gross supply of mortgages should be sustained in the near term, there are a couple of very important mitigating factors.
First, some of the largest servicers have informed us that over the last six months the multipliers they use to book servicing have decreased from around 7 to 5. This was the result of losses from prepayments, lower Trust IO prices and the decreasing value of the float.
A steep curve generally lowers the value of servicing (opposite as for Trust IOs) by lowering the value of the carry earned between payment collection and distribution. That cost will likely be passed on to the consumer, leading to mortgage rates between 10 to 15 bps higher than would otherwise have been the case.
It is also important to note that strong rolls in the mortgage market also impact the primary market mortgage rates. As most transactions take place for 30- to 60-days forward settle, the rates will be 8 to 16 bps higher due to the rolls. The steep curve has therefore impacted the primary rates in two ways: higher servicing costs and higher forward mortgage rates.
Net supply to decrease 30%
in the next two quarters
We expect that net MBS issuance will actually drop in the coming months. Historical data indicates that net supply is primarily driven by home-price appreciation, with no additional correlation to interest rates. We find that there is approximately an eight-month lag between net supply and home price appreciation. The correlation between the two has been around 80% for the last 10-years.
Based on recent home price appreciation through the second quarter of 2001, we estimate a 30% decline in net supply by the beginning of 2002. If we extrapolate further to a major recession and 2-3% home-price appreciation levels (matching 1992/93), then our projection for net production is a 67% decline from current levels by the end of 2002. That would imply about $10 billion in net production. Given bank and GSE purchases, we clearly expect a significant tightening in MBS spreads, despite relatively large gross supply.