The scheduled completion of the Federal Reserve's MBS purchase program this March has created fears that rising mortgage rates will exert renewed pressure on home prices. While the program has been successful in keeping primary rates low, it has had some unanticipated effects that will impact the housing and capital markets even after outright purchases have ended.
The Fed's $1 trillion or so of net purchases last year were concentrated in 30-year conventional coupons, with FNMAs comprising roughly 57% of the total. Their conventional purchases have ranged across the coupon stack; 4s and 4.5s comprised 56% of their net buying, with the remainder concentrated in 5s and 5.5s.
By spreading their purchases across the coupon stack, the Fed avoided the types of distortions that would have resulted from limiting their activity to current coupons. However, their net purchases had a major impact on market liquidity, particularly for non-production coupons. For example, the net supply (i.e., new issuance less runoff) of FNMA 5.5s was negative $119 billion for all of 2009, reflecting both limited new issuance and relatively fast prepayment speeds for the in-the-money coupon. The Fed's purchase of an additional $150 billion of the coupon means that a total of $268 billion in FNMA 5.5s were removed from the market last year.
This has clearly impacted liquidity in conventional MBS with 5% and higher coupons, all of which had negative net supply after accounting for the Fed's activities. This has manifested itself in both very tight dollar roll markets (with 5s and 5.5s consistently trading to negative costs of funds after Labor Day), and talk of large and persistent fails.
As a result, the Fed was obliged to aggressively sell dollar rolls for 5s and 5.5s last fall, despite the fact that this fund-draining activity is contrary to the goal of "quantitative easing." With conforming mortgage rates remaining close to historical lows, there is little prospect of additional supply in higher coupons. This suggests that liquidity issues in these coupons will persist, and may force the Fed to continue trading dollar rolls even after the outright purchases of MBS end.
However, the MBS purchase program improved the risk profile of the fixed-income markets by transferring risk to the Fed's balance sheet. Using Yield Book effective durations, I estimated that the Fed pulled the equivalent of about $660 billion in 10-year Treasurys out of the MBS market last year. This ultimately dampened the duration swings resulting from market moves and reduced the volume of hedging trades necessary for investors to collectively maintain portfolio durations. The analysis suggests that $44 billion fewer 10-years would need to be sold in a 25-basis-point rise in rates, while roughly $85 billion less 10-years would need to be purchased in a comparable rally.
Interestingly, the reduction in the negative convexity of the fixed income markets was associated with declining levels of realized volatility for intermediate-maturity Treasurys over the course of last year. The 60-day standard deviation of daily changes in the 10-year Treasury's yield declined from around 10 basis points in July to less than six basis points by the end of 2009.
In this light, the greater stability of intermediate yields was a second-order effect of the purchase program, and probably contributed at the margin to the improved valuations of MBS last year. Moreover, lower levels of realized volatility are likely to persist after the purchases end, as it may take years for issuance to replace the mortgage duration removed from the market. Therefore, some of the program's benefits to the MBS and mortgage markets may continue even after the program is terminated.
Bill Berliner is a mortgage and capital markets consultant based in Southern California. His Web site is www.berlinerconsulting.net.
(c) 2009 Asset Securitization Report and SourceMedia, Inc. All Rights Reserved.