In a recent post, we discussed using Recursion’s proprietary tools to unravel the Federal Reserve’s MBS holdings of Fannie Mae and Freddie Mac loans. The Fed’s holdings, however, are part of a bigger picture issue regarding the notion of “float” in the MBS market, that is, the amount of securities outstanding that are available to trade. The holdings of the central bank serve to reduce the float as the Fed is a buy-and-hold investor. These loans are said to be “locked up”. Besides the Fed, loans can be locked up in structured products, notably Collateralized Mortgage Obligations (CMOs). The first CMOs were launched by Freddie Mac as Real Estate Mortgage Investment Conduits (REMICs) in 1988 and allow cash flows to be tranched to meet the needs of different investors. Pools in CMOs’ collateral groups are also locked up.
In a recent paper, researchers at the Philadelphia Fed (Liu, Song and Vickery, May 2021) discussed the history of the differential between the pricing and the trading volume differential between Fannie Mae and Freddie Mac securities. Historically, Fannie Mae securities had traded ten times as often as those of Freddie Mac, with the consequence that trading costs for Freddie Mac could be twice those of Fannie Mae. In this paper they comment that Freddie Mac compensated for this by raising its g-fee, and by locking up its securities in CMOs.
Using the same recursive algorithms as in our prior blog, we can back out the CMO lockup, FED lockup and Float by agency:
With talk of taper at the top of the monetary policy discussion, it is worthwhile to dig a bit into the role of the Federal Reserve in the functioning of the MBS market. As is well known, the onset of the Covid-19 pandemic resulted in a resurgence of central bank purchases of Agency mortgage-backed securities (MBS).
We received the monthly GSE data download for the June book of business over the weekend and prepayment speeds dropped for the second consecutive month, with the 1-month CPR printed 22.4, the low posted since 17.1% was reached in February 2020 just before the onset of the pandemic.
Mortgage rates are of course the key driver here, but other issues matter as well, notably lending capacity. With the onset of the pandemic and the associated loosening of monetary policy and spike in demand for housing away from dense locations, the mortgage industry became overwhelmed. Originators were busy hiring and increased their capacity over the past 18 month to deal with the long period of refinancing activity. However, as prepayment speeds slow down, it appears that the capacity building may be overshooting. In response, originators have started to lower their underwriting standards to create enough volume to fully utilize the capacity.
Traditionally, the industry fine-tunes its production through tweaking its credit standards to keep its pipeline as full as possible. This is occurring now notably for refinance mortgages:
What we can see is that purchase demand remains strong, with the swing product being refinance mortgages. It is evident that lenders are trying to smooth out refinance production with countercyclical credit tightening and loosening. As credit scores are higher than was the case in the pre-pandemic period there is room to ease further, but the ultimate extent is highly uncertain.
As we have commented several times, the Federal Reserve Z.1 data is a fine source of information on long-term financial market trends. This post looks at trends in ownership of single-family mortgage risk. The chart below shows this distribution from Q1 2007 to Q3 2020:
The role of the Federal Reserve in the mortgage market is an ongoing theme in this blog, dating back to the early days of the Covid-19 crisis. Besides the level of short-term rates, when we think about the Fed, we consider its direct intervention through the outright purchases of MBS. A common view of this “QE” policy can be seen in charts such as:
In a previous post, we noted that a key component of the implementation of unconventional monetary policy is the selection of coupons in its MBS purchase programs. As coupons in these securities occur only in steps of 0.5%, obtaining liquidity in new lower coupons is important in establishing a basis for mortgage rates to move lower. For example, 2.5% 30-year coupons were issued in 2013 and 2016, but never to the extent that liquidity was firmly established, limiting declines in the rate that borrowers paid at those times. This changed dramatically with the onset of the Covid-19 crisis. In March 2020 the Fed restarted its MBS purchase program after six years, including the 2.5% coupon, and this soon became the dominant coupon.
Since that time, the Fed has picked up its activity on this front. The GSE’s started issuing 2.0’s shortly after the crisis hit, and the Fed began buying them in May. Then 1.5% came on the scene this summer and this week sure enough the central bank validated this activity by adding them to their portfolios.
Can 1.0’s be far behind?
The Federal Reserve recently released its quarterly Z.1 report: The Financial Accounts of the US (formerly known as the Flow of Funds) for Q2 2020. This voluminous dataset contains very detailed information describing financial flows and stocks across all major segments of the economy. For our purposes the key chart is the distribution of holdings of Total Long-Term Agency Debt (Agency MBS + Agency notes and bonds):
The clear takeaway for Q2 is the central bank gained share as it took unprecedented actions to stabilize the financial system in the wake of the Covid-19 crisis. The Fed’s share of the stock of Long-Term Agency Debt jumped by about 4.9% from Q1 to Q2, a record high increase. The biggest losing category was “Others” which fell by 3.6%. This loss came largely from the household subsector within this category. The “Commercial Banks” share rose by 0.9%, but this does not necessarily imply greater appetite for mortgage risk as depositories are increasing their securitization rate by swapping whole loans for securities. Finally, the “Foreign Investor” category lost 0.5% to a 7-quarter low.