Case Study: Mortgage Hold vs. Sell Analysis

Hi, I’m Tyler Buckridge with the Federal Home Loan Bank of Boston.

In this case study, I want to focus on a simple idea.

A mortgage coupon is not the same thing as mortgage value.

A higher coupon can make new residential mortgage production look attractive to hold, but the hold decision still depends on expected life, borrower optionality, funding cost, retained risk, balance sheet capacity, and what else the institution could do with the cash.

This case study goes deeper into that hold versus sell decision.

For a broader introduction into the Mortgage Partnership Finance Program and how it works, see Caroline Casavan’s article, Should You Participate in the Mortgage Partnership Finance Program, which is available on our website.

Here, we are focusing on the balance-sheet economics of selling new production through MPF 35.

We are going to take a sample pool and ask a practical ALCO question.

Should the member hold the loans in portfolio?

Should the member sell them through MPF 35, or should the member use a partial sale strategy? The key point is holding should be a decision, not the default.

I’ll cover three things. First, the decision framework.

We will look at expected life, all-in hold value, balance sheet capacity, risk transfer, and the use of sale proceeds.

Second, we will look at the MPF 35 mechanics needed to make a comparison.

That includes the transfer of interest rate and prepayment risk, the retained credit enhancement position, servicing, and the choice between upfront and regular execution.

Third, we’ll put the numbers together.

The model compares the five-year value of holding the pool with the value of selling through MPF 35 and redeploying the proceeds.

The goal is not to argue that every mortgage should be sold.

The goal is to identify the economic hurdle that a mortgage pool should meet before it earns a permanent place on the balance sheet.

This slide is the decision lens for the case study.

The first question is expected life. How long is the coupon likely to remain outstanding?

The second question is all-in hold value.

What does the member earn after funding cost, servicing and administrative expense, credit cost, and the effect of principal runoff?

The third question is balance sheet capacity.

Would holding the pool consume liquidity, add to residential concentration, or use capacity that could support another priority.

The fourth question is risk transfer.

How much interest rate, prepayment, and credit exposure does the member retain under each alternative?

And the fifth question is use of proceeds.

What could the balance sheet do with the cash if the pool were sold?

That can include liquidity, liability reduction, securities, new loan growth, or a combination of uses.

The hold decision, therefore, begins with expected life and opportunity cost, not stated yield alone.

This table shows what changes when a member sells through MPF 35 instead of holding the loans.

If the member holds the pool, it retains the interest rate risk, prepayment risk, full hold loan credit exposure, and the liquidity impact of keeping the assets on balance sheet.

If the member sells through MPF 35, the interest rate and prepayment risks are transferred.

That matters because the borrower controls the refinance option.

When rates fall, the borrower can refinance.

The institution cannot require the higher-yielding mortgage to remain outstanding. Also, credit risk is not eliminated.

Instead, full whole loan exposure is converted into a defined position within the NPF loss waterfall. Servicing can be retained or released.

This case study assumes retained servicing, so the member preserves the customer relationship and servicing income.

And selling creates cash that can be directed towards another balance sheet need. So NPF changes more than income.

It changes the risk package, liquidity position, and future flexibility of the balance sheet.

This slide shows where the retained credit exposure sits.

Losses are absorbed in layers. Borrow equity comes first.

Then, if private mortgage insurance is present, it also provides protection ahead of the FHL Bank Boston and PFI layers.

Next comes the FHL Bank Boston first loss account.

Only after those layers are exhausted does the participating financial institution credit enhancement obligation begin absorbing losses.

The participating financial institution, or PFI, is the member selling the loans through MPF. Losses above the PFI obligation move to the FHL Bank Boston’s catastrophic loss layer.

The important distinction is that MPF 35 does not create zero credit risk, but the member is also not retaining the same exposure it would have by owning the entire mortgage pool.

It retains a defined credit layer while transferring the interest rate and prepayment risks.

While with that structure established, we can now apply the framework to a specific pool.

The sample member is a hypothetical $1 billion depository evaluating $10 million of newly originated 30-year fixed-rate residential mortgages.

The member has a 90% loan-to-deposit ratio, borrowings equal 10% of assets, one-to-four-family loans equal 50% of assets, securities equal 10% of assets, cost of funds is 2%, and the net interest margin is 2.5%.

This pool contains 25 loans, and the average loan size is $400,000.

The note rate is 6.5%, and servicing is retained.

This is not a distressed institution, but it is a balance sheet where capacity has value.

Loan to deposit is already high, wholesale funding is meaningful, half of assets are already in one- to four-family mortgages, and the securities book is relatively limited.

That makes the next 10 million of production more than an asset selection decision.

It is also a funding, liquidity, concentration, and opportunity cost decision.

The question is whether this pool creates more value by remaining on balance sheet or by being sold through MPF 35 and redeployed.

Now we can size the retained credit position for this pool. The total credit enhancement output is $357,399 or 3.57% of the $10 million pool.

The FHL Bank Boston first loss account is $35,000 or 35 basis points, hence the name MPF 35.

That leaves an implied PFI credit enhancement obligation of $322,399.

The PFI obligation does not begin with the first dollar of mortgage loss, however.

It attaches after borrower equity, any applicable mortgage insurance, and the first loss account.

The NPF credit enhancement estimator already incorporates the submitted loan characteristics, which include the credit profile, loan-to-value, and mortgage insurance, So, those protections do not need to be factored in a second time.

The other important point is how the obligation enters the economics.

The $322,399 is not treated as a day one cash expense.

It is a retained credit exposure and capital consideration.

The model, therefore, applies a risk or capital charge to that exposure rather than subtracting the entire obligation from day one proceeds.

That allows the analysis to recognize the retained risk without overstating it as an immediate cash outflow. Credit enhancement is not ignored.

It is sized, compensated through credit enhancement income, and explicitly carried in the comparison. The next decision is how to execute the sale.

The regular MPF 35 option in this example has a day-one premium of approximately 1.28% or about $128,000. The upfront option has a premium of approximately 1.53%, or $153,000.

So the upfront option brings an additional 25 basis points, or roughly $25,000, into day one.

The trade-off is recurring credit enhancement income.

The regular option assumes seven basis points in year one and 14 basis points after year one, subject to performance.

The upfront option assumes no credit enhancement income in year 1 and 7 basis points after year 1.

Both cases assume the member retains servicing and receives a 25-basis point servicing fee on the outstanding balance.

So regular versus upfront is fundamentally a timing decision.

A longer-lived pool gives recurring income more time to compound.

A faster prepaying pool increases the value of capturing more economics at delivery.

Expected life therefore affects not only whether to sell but also how to sell.

This is where the hold case begins to separate from the stated coupon.

The member earns 6.5% only on principal that remains outstanding.

At a 5% conditional prepayment rate, approximately 6.8 million of the original pool remains after five years.

At 10% CPR, the remaining balance is about 4.8 million. At 13.9% CPR, only about 3.5 million remains.

That 13.9% assumption is the 12-month CPR for the 6.5% coupon bucket in the May 2026 NPF data.

At that pace, roughly two-thirds of the original pool is gone within five years.

At 25% CPR, the remaining balance is close to one million.

So a 6.5% coupon on a $10 million pool is not economically equivalent to earning 6.5 percent on 10 million for five years.

The income is earned on a declining balance, and the speed of that decline is controlled largely by borrower behavior.

That is the embedded option in the hold case.

If rates fall, the mortgages that look most valuable to retain may also become the mortgages most likely to leave.

The relevant measure is therefore not stated yield by itself; it is yield multiplied by expected life.

Selling creates a different earnings path.

Instead of earning mortgage income on a balance that amortizes and prepays, the member receives the sale proceeds and can direct that cash towards another balance sheet objective.

The table uses illustrative rates and avoids cost assumptions, not executable market quotes.

Liquidity is modeled at 3.5%.

Liability reduction is modeled across the 4% to 5% range.

Treasuries and agencies are shown at 4.25%.

MBS and CMOs are shown at 5%.

New loan growth is shown at 5.75%.

And blended redeployment is shown at 4.75%.

The economic value of those choices is not identical.

Liquidity may have a lower direct yield, but greater contingency value.

Liability reduction produces an avoided funding cost.

Securities provide liquid earning assets.

New loan growth may produce a higher return and preserve capacity for priority customer relationships.

The sale, therefore, is not the strategy by itself. The strategy is what the balance sheet should do next.

A member-specific analysis would replace these illustrative assumptions with the institution’s actual marginal funding cost, liquidity value, reinvestment opportunities, and expected loan returns.

This slide brings the full comparison together.

The hold case produces approximately $790,000 of modeled five-year net contribution.

That result reflects mortgage income on the declining outstanding balance net of the modeled funding, servicing, credit, and other carrying assumptions.

The NPF alternatives are built differently.

They include the day one sale premium, retained servicing income, credit enhancement income, and five years of return or avoided cost on the sale proceeds less the model charge for the retained credit enhancement exposure.

The member also transfers the interest rate and prepayment risks.

The largest driver in the difference of outcomes here is straightforward.

The hold case earns its spread on a balance that amortizes and prepays. The sale case puts nearly the full proceeds to work immediately.

Under upfront MPF 35 produces about $2.49 million. Those two outcomes are only about $7,600 apart.

And that’s useful because it shows that the regular versus upfront delivery decision is secondary to the larger hold versus sell decision under these assumptions.

The extra upfront premium is largely offset over time by the regular option’s higher recurring credit enhancement income. Upfront MPF 35 with MBS or CMO reinvestment produces about $2.78 million.

Upfront MPF 35, with new loan growth, produces about $3.16 million.

And upfront MPF 35 with blended redeployment produces about $2.65 million.

Compared with the $790,000 hold result, the modeled MPF advantage ranges from approximately 1.69 million to 2.73 million dollars.

The result is large because the comparison is not coupon versus sale price.

It is the declining economics of the held pool versus the complete economics of sale, retained income, risk transfer, and redeployment.

The heat map on this slide tests whether that conclusion survives different assumptions.

Each cell shows MPF 35 value minus hold value over five years. Zero is breakeven.

A positive number favors MPF 35, and a negative number would favor holding. Every scenario shown is positive.

At a 5% CPR and 3.5% liquidity return, the modeled MPF advantage is still $1.02 million.

At the 13.9% CPR reference point, the advantage ranges from about $1.22 million to $2.75 million, depending on the use of proceeds.

At 35% CPR and a 6.5% redeployment return, the advantage is approximately 3.7 million dollars.

So for this sample pool, sale beats hold across every CPR and redeployment combination displayed.

But that does not mean every mortgage should be sold.

It means this pool does not clear the economic hurdle required to justify holding under the assumptions tested.

Holding would become more competitive if expected life were materially longer, the member had a durable low-cost funding advantage, liquidity and residential concentration capacity were abundant, MPF execution were weaker, or if the member had little productive use for the sale proceeds. Those are the conditions that would move the model towards breakeven or towards hold.

But in this sample, faster runoff and productive redeployment consistently shift the answer towards MPF 35.

The practical next step is to run the framework before the next production pool becomes a permanent balance sheet position.

The first input is the pool.

We need the expected balance, coupon, product type, servicing approach, credit profile, and the production volume.

The second input is the MPF 35 execution.

That includes current pricing, credit enhancement output, retained servicing income, credit enhancement income, and the retained obligation.

The third input is the member’s balance sheet alternative.

What is the actual marginal funding cost? How valuable is additional liquidity?

Is the priority liability reduction, securities, customer loan growth, or a blended redeployment strategy? We can then compare three distinct choices.

Hold the production, sell the production through MPF 35, or use a partial sale strategy that retains some mortgages and sells the rest.

The answer may differ by member and may change as rates, funding needs, and loan demand change, but the analysis should happen before holding becomes automatic.

For members originating new residential mortgages, FHL Bank Boston’s strategies, MPF, and relationship management teams can apply this framework to the next production pool using current execution and the members’ actual balance sheet assumptions.

Let’s land the plane.

For this sample pool, MPF 35 wins because the combined economics of sale-execution, retained income, and redeployment exceed the economics of earning a 6.5% coupon on a declining and potentially short-lived balance.

The member also transfers the interest rate and prepayment risks. That is the answer to the case study. It is not that every mortgage belongs off balance sheet.

It is that every new pool should clear the full economic hurdle before it is held.

So before the next pool of production becomes the default hold decision, bring us the assumptions.

We can compare hold, sell through MPF 35, and a partial sales strategy using the member’s actual loans, current MPF pricing, credit enhancement output, funding costs, and use of proceeds. The next mortgage should earn its place on the balance sheet.

Thank you for taking the time to go through the case study, and if you’d like to start analysis on your next pool of production, please reach out to myself or your relationship manager.