The following are four approaches one can take to efficiently hedge the price risk of holding residential mortgage loans prior to bulk sale, improving execution and profitability. Not surprisingly, they intersect and are progressively more robust.
Method 1. Forward Sales
The first method is to sell the loans forward, similar to what occurs in the Agency To-Be-Announced (TBA) market, where TBAs are sold today and eligible loans are delivered during the settlement window in the future. Although there is some forward selling occurring in the Non-QM market, capital markets professionals are not observing meaningful price improvement and find it having limited use. The process is also bi-lateral, lacking uniformity and liquidity. However, it is an option, and it does transfer all risk - but at the expense of potential price improvements from hedging the interest rate risk and seeking better execution through bulk sales.
Method 2. Correlated Hedges
The second approach is a data-based regression method, involving the selling (short selling) of US Treasuries, or paying fixed on SOFR swaps, as these are both correlated with the price of loans. However, regression analysis requires good quality loan price data to analyze against US Treasury or SOFR swap prices. While the US Treasury and SOFR swap markets are both liquid and transparent, offering accurate and available data, this is not the case for Non-QM loan data. Challenges include:
- Data Quality: Due to the shortage of accurate and uniform baseline Non-QM loan price data, regression results can have limited use.
- Theory Flaws: Co-linearity, for example the correlation of 10-year rates with other tenor rates (e.g. 2, 4, 5 and 7-year rates), may erroneously suggest a hedge instrument that does not match the expected maturity of the loans being hedged.
- Practicality: Co-linearity may also recommend impractical hedges, such as a butterfly strategy of short 3-year, long 5-year, and short 10-year SOFR swaps to hedge a loan pool that would be more accurately hedged by shorting all 3 of these maturities.
Method 3. Hedging to Expected Prepayments (CPRs)
The third method involves hedges that match the expected prepayment schedule of the loan pool. This effectively hedges the funding interest rate risk that would be required to hold the loans to final maturity. Conditional Prepayment Rates (CPRs), widely used to explain borrower prepayment behavior, can inform us of the outstanding unpaid principal balance (UPB) of a loan pool, and hedges are then established to best align with this profile.
Hedge Example: A Non-QM loan pool with an average market-neutral coupon may have an 18% CPR, implying a weighted average life of 4.5–5.5 years. One approach might be to hedge to the expected 5-year average life, shorting the 5Y Eris SOFR Swap futures contract. However, the loan pool is not a bullet repayment loan; it prepays steadily over time. Therefore, a more accurate approach would be to utilize a combination of 2, 5, 7 and 10-year Eris SOFR Swap futures to hedge the funding interest rate risk more precisely with the expected prepayment schedule.
Method 4. Hedging to a Stochastic Model Price
The fourth and most robust, but more complex approach to pricing and hedging Non-QM loans is to use a stochastic, discounted cash flow (DCF) model. Stochastic is a fancy word used in probability modeling, meaning that interest rates will move in a random manner over time according to a probability distribution implied by market traded volatility, rather than following the risk-free (i.e.hedgeable) future path of rates. The model determines expected future cash flows based on probabilities of prepayment, and then discounts the expected cash flows to today.
In evaluating the financial instruments on which to build this model, one quickly will find that it is not possible to trade future primary mortgage rates to inform us of future borrower prepayment behavior. Therefore, models are based on SOFR swaps, the most efficient way to model, trade and hedge the future path of interest rates. Once the future path of SOFR rates is determined, a mortgage-SOFR spread is added to model the future path of mortgage rates and therefore borrower behavior.
The Case for SOFR Swaps vs. US Treasuries
Financial modelers prefer using SOFR swaps to build their models because they are liquid, easily accessible, and efficient to trade. Critics may ask: "Why not use US Treasuries?" The reason is twofold. First, it is inefficient to trade forward expectations of US Treasury rates as there is no efficient market in term financing of US Treasuries beyond a few months. This is necessary to price and hedge forward US Treasury rates. This is the balance sheet effect, with US Treasury securities requiring the financing of the full purchase price, while SOFR swaps are derivatives, requiring only the financing of initial margin. Second, credit idiosyncrasies of US Treasuries muddy the water when one is hedging forward mortgage rates. Therefore SOFR swaps, liquid and easily accessible as Eris SOFR Swap futures, are favored in running models and trading SOFR swap hedges.
Conclusions
All methods outlined above intersect. However, the limited availability of Non-QM loan price data means that determining hedges by regression of poor quality Non-QM loan price data with high quality US Treasury or SOFR swap data, can yield inaccurate hedges. Consequently, replicating the prepayment profile of loan pools with more precise Eris SOFR Swap futures hedges - Method 3 - or the use of a robust stochastic model based on SOFR swaps - Method 4 - are today’s favored approaches to hedging Non-QM and improving execution and profitability.
To learn more about hedging Non-QM with Eris SOFR, contact John.Douglas@erisfutures.com.
Date
