Uncategorized
Evaluating risk-adjusted returns across mortgage credit
admin | September 11, 2020
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
In the wake of heightened market volatility across all sectors of mortgage credit this year, investors are likely evaluating performance in terms of not only nominal returns but the relative stability of those returns. Mortgage credit portfolios have seen extremely elevated volatility this year, more so than at any other point in recent memory absent market dislocations in the wake of the Global Financial Crisis. Given this volatility, one metric for evaluating the recent performance of mortgage credit is by looking at profiles that have offered the best return with the least amount of volatility.
Based on this metric, bonds at the top of the capital structure in re-performing loan securitizations and GSE Credit Risk Transfer deals have outperformed other exposures to mortgage credit. These bonds have even outperformed investment grade corporate credit despite the significant disadvantage of being one of the few domestic fixed income asset classes without explicit support from the Federal Reserve. RPL and CRT senior bonds provided lower average nominal returns than the broader investment grade corporate index over the past year but showed significantly lower volatility of those returns. Given this, investors may look to lever these cash flows to replicate the volatility of an investment grade corporate benchmark and enhance returns over that benchmark. Risk-adjusted returns are far less favorable across the capital structure in legacy RMBS and in mezzanine and subordinate RPL and CRT bonds which almost unilaterally underperformed high yield credit.
Prior to March, mortgage credit assets have historically provided an attractive source of out-of-index exposure, offering investors both alpha and attractive risk-adjusted returns, largely driven by favorable technical and fundamental tailwinds. Those steady returns came to an abrupt halt in March as some areas of mortgage credit posted negative returns of more than 50%. Negative returns driven by falling asset prices have since reversed rapidly but have left a trail of volatile price swings in their wake and given this, investors may look to pivot to areas of the market that have provided more stable returns, especially given uncertainty around the lingering impact of Covid-19-related losses especially further down in the capital structure.
Sharpe ratios and risk equivalents at the top of the stack
The analysis focuses on three sectors of mortgage credit. GSE CRT, newly securitized re-performing loans and legacy RMBS and employs a relatively simple taxonomy of senior, mezzanine and subordinate risk. Obviously these risks are not all the same in the different expression of mortgage credit as CRT by definition has no true senior bond. However for the purpose of the analysis, M1 classes of CRT will serve as a proxy for the senior portion of the capital structure. Even comparing bonds at the top of the defined capital structure has some limitations as bonds at the top of the stack in CRT and RPL deals exhibit significantly less spread duration than most legacy seniors given the sequential nature of those structures, dampening price volatility in these assets relative to equal spread widening in legacy seniors. Evan after making concessions for spread duration the difference in risk-adjusted returns between CRT/RPL and legacy bonds are pronounced. This is somewhat surprising given the greater price transparency in GSE CRT bonds relative to other forms of mortgage credit although the effect of this may be somewhat muted by the fact that the analysis only observes month over month price changes and not intra-month volatility . (Exhibit 1)
Exhibit 1: Risk-adjusted returns at the top of the stack
Monthly total return calculated as monthly price change, interest income, and discount par principal accretion net of principal loss. RPL seniors defined as original AAA bonds only. 12 month observation on legacy RMBS is shifted one month (August 2019-July 2020) Source: IDC, Amherst Insight Labs, Amherst Pierpont Securities
The relatively low volatility off return evident in shorter, higher rated cash flows potentially allows for investors to employ financial leverage to replicate the return volatility of a benchmark portfolio, effectively creating a ‘risk equivalent’ portfolio. For example, over the past 12 months the Citi Broad Investment Grade unsecured corporate bond index has had an average monthly return of 64 bp and a standard deviation of that return of 0.0301 while CRT M1 classes have had an average monthly return of 23 bp but a standard deviation of just 0.0064. To the extent the financing terms and availability allow, investors can lever CRT M1s 4.7 times to create a portfolio that replicates the risk profile of the investment grade index but offers an incremental 45 bp of excess return over the benchmark. (Exhibit 2)
Exhibit 2: Building a framework of risk equivalents
Monthly total return calculated as monthly price change, interest income, and discount par principal accretion net of principal loss. RPL seniors defined as original AAA bonds only. 12 month observation on legacy RMBS is shifted one month (August 2019-July 2020) Source: IDC, Amherst Insight Labs, Amherst Pierpont Securities
Risk and return in mezzanine and subordinate credit
Moving further down the capital structure, risk-adjusted returns over the past year appear both non-intuitive and unlikely to be replicated as subordinate classes of GSE CRT offered modestly better risk-adjusted returns than RPL and prime and Alt-A legacy mezzanine bonds. CRT B2s offered the highest risk-adjusted returns of bonds analyzed. CRT B1 and B2 classes both produced negative returns of greater than 50% in March and subsequently posted cumulative returns of 63.6% and 78.1% respectively in the second quarter of this year. With a Sharpe ratio of 0.34, CRT B2s have an almost identical risk-adjusted return as the Citi High Yield index which had a 12-month Sharpe Ratio of 0.33. However the volatility of returns in CRT B2s was roughly seven times that of the high yield index. Somewhat surprisingly, CRT last cash flows exhibited some of the least attractive risk-adjusted returns of mezzanine and subordinates analyzed as they posted lower average nominal returns and higher volatility of returns than RPL or legacy mezzanine bonds. As stated earlier, some of this may be attributable to the greater market pricing transparency evident in CRT, but given the fact that the volatility is measured simply as month-over-month price change, some of that impact should be negated. (Exhibit 3)
Exhibit 3: Risk and return across mezzanine and subordinate mortgage credit
Monthly total return calculated as monthly price change, interest income, and discount par principal accretion net of principal loss. RPL seniors defined as original AAA bonds only. 12 month observation on legacy RMBS is shifted one month (August 2019-July 2020) Source: IDC, Amherst Insight Labs, Amherst Pierpont Securities
In this instance, past is not likely prologue as it appears unlikely that the market will experience another technical dislocation that drove a rapid whipsaw in valuations across various levered forms of mortgage credit. The net result of the events of March was an overall reduction in financial leverage employed by holders of levered mortgage credit assets. Ultimately, if a comparable dislocation in deep mortgage credit were to occur again it would likely be more fundamental in nature and as such, there would likely not be such a profound recovery in asset valuations.