Peakhill: What the Watchlist Really Says About the CMHC Exit Path

The Point Of This Follow-Up

The main article argued that Peakhill is not really living on loan yield alone. It is living on the ability to turn short bridge loans into permanent CMHC financing. This follow-up isolates the exact place where that claim is tested: the watchlist, the arrears bucket, and the movement between Stage 1, Stage 2, and Stage 3.

There is one especially unusual finding here. At the end of 2025, the watchlist stood at CAD 56.276 million. At the same date, Stage 2 exposure was CAD 29.119 million and Stage 3 exposure was CAD 27.157 million. The sum is identical. In other words, by year-end the watchlist is not just an operating list used by the asset-management team. In practice it matches the entire portion of the book that has already left Stage 1.

That matters even more because the model is built around one dominant exit channel. Peakhill says that more than 90% of its loans are repaid through CMHC refinancing. In the current book, about 71.3% are CMHC-eligible bridge loans, about 5.6% are construction loans already insured by CMHC, and another 4.1% are cash-flowing multifamily loans that are also CMHC insured. Put differently, roughly 81% of the portfolio depends directly on the availability, continuity, or eligibility of that same exit route.

The short version up front: the risk does not begin and end with the 9 loans already in Power of Sale. Almost half of the watchlist is already sitting in Stage 3, but the other half, CAD 29.1 million, is still sitting in the middle zone. So the 2026 test is not only whether the defaulted loans are recovered in full. It is also whether the middle layer exits back through CMHC before it rolls forward into Stage 3.

The Watchlist Here Is An Early-Warning Layer, Not A Routine Operating Note

Peakhill's default definition looks fairly strict on paper. The company treats a mortgage as a default event if the borrower is more than 90 days past due on multifamily loans, more than 30 days past due on other mortgage types, or if another breach occurs, such as unpaid property taxes or a material adverse change in the borrower or the collateral. At the same time, as of the report date there were 9 loans totaling about CAD 28 million already in Power of Sale, with management expecting full recovery through asset realization within 45 to 90 days, average loan to value of 75% to 80%, and personal guarantees on all of them.

But the watchlist is clearly wider than that formal default definition. The company says a loan can enter the watchlist because of covenant concerns, drift from the original underwriting case, credit or valuation concerns, weakening property performance, property-tax or insurance issues, borrower developments, reporting issues, or simply approaching maturity. It also says that its weekly review process looks at loans with more than four NSF payment returns in the prior year. This is not only a collections list. It is an early-warning system for deals whose exit path is no longer clean.

The more important point is that the company explicitly says that the loans shown on the watchlist, all or part of them, are also loans exposed to the risk that guaranteed permanent CMHC financing will not be obtained. So the watchlist is not just a collections indicator. It is an early map of exit-path friction.

This is also where one important caution sits inside the historical record. The company says there were no delinquencies across the life of the portfolio, but in the same section it explains that payment delays do exist and are not treated as a loan default unless more than 90 days have passed. The reasonable reading is that the apparently clean historical line does not necessarily mean a frictionless past. It means friction that was handled before it crossed the threshold that turns it into a formal default statistic.

By Year-End 2025 The Entire Watchlist Was Already Outside Stage 1

The full overlap between the watchlist and the sum of Stage 2 plus Stage 3 is the most important insight in this continuation. If you add CAD 29.119 million of Stage 2 and CAD 27.157 million of Stage 3, you get exactly CAD 56.276 million, the entire watchlist. That strongly suggests that at the year-end cut there is almost no gap between the operational monitoring layer and the accounting staging layer. Everything requiring intensified oversight has already left the bucket considered fully performing.

The internal split of that list matters just as much. About 48.3% of it is already in Stage 3, but about 51.7% is still in Stage 2. Anyone looking only at the CAD 27.2 million already classified in default misses the other half of the story. Nearly the same amount of credit has already left Stage 1 without yet reaching formal workout. That is exactly the zone where the CMHC exit path is supposed to prove whether it can arrest deterioration rather than simply manage it after default has already arrived.

The stage-movement table sharpens another non-obvious point. In 2025, CAD 47.07 million moved out of Stage 1. In the same year, CAD 25.23 million moved into Stage 2 and CAD 21.84 million moved into Stage 3. The sum again is nearly exact. That suggests the year’s deterioration was not created only by a bigger book. It also came from a real shift of loans out of the performing layer and into the monitoring and default layers.

This is the key point: the portfolio did grow very quickly, with CAD 588.058 million of mortgage advances in 2025. But underneath that growth, almost CAD 47 million changed layer for the worse. Anyone reading 2025 only as a volume story will miss the internal credit migration.

Provisioning Did Not Panic The Way Stage 3 Did, Because Management Still Assumes Recovery

At first glance the allowance can look conservative enough. Total mortgage-loss allowance rose from CAD 3.0 million to CAD 6.59 million. But when the number is opened up, the story is different. Most of the increase came from Stage 1, where the allowance rose from CAD 2.556 million to CAD 5.84 million as the book expanded. In Stage 2, allowance increased from CAD 23 thousand to CAD 334 thousand. In Stage 3, however, allowance barely moved at all, from CAD 421 thousand in 2024 to CAD 416 thousand in 2025, even as the exposure itself rose from CAD 10.919 million to CAD 27.157 million.

That is exactly why Stage 3 coverage fell from 3.86% to 1.53%. The company explains that each Stage 3 mortgage is assessed individually based on expected recoveries, and only the expected shortfall is provisioned. The inference from the disclosed numbers is that management still reads most of the problem as an event of delayed exit, realization, and time, rather than a broad principal-loss event.

That reading has clear anchors in the filing. Across the life of the mortgage portfolio, the company reports 14 default cases totaling CAD 30.858 million, and says all of that amount was fully recovered. It also says there have been no realized losses across the life of the portfolio. On top of that, management says the 9 loans already in default at the end of 2025, totaling roughly CAD 27 million to CAD 28 million, are also expected to be recovered in full through asset realization within 45 to 90 days, supported by average 75% to 80% LTV and personal guarantees.

But this is exactly where 2026 becomes a harder test. The current default bucket is almost as large as the entire cumulative historical default volume the company has disclosed so far. So even if the historical recovery pattern ultimately repeats, the system is now being asked to digest, at one reporting date, almost all of the historical default volume at once. On top of that, 2025 expected-credit-loss expense also included a one-time write-off of CAD 408 thousand of interest receivable in arrears. So the stress has already reached earnings, even if it has not yet shown up mainly as realized principal loss.

What Has To Close From Here

From here, the question is not whether Peakhill can keep originating loans. It already proved that. The question is whether the CAD 29.119 million in Stage 2 can return to an exit path, or whether it becomes tomorrow’s Stage 3. That is no longer a theoretical issue, because 86.4% of the entire portfolio matures within 12 months. In a structure like that, an enlarged watchlist is not background noise. It is an operating bottleneck.

The second check sits against management’s own timing claim. If the 9 loans already in Power of Sale are in fact recovered within 45 to 90 days and without material loss, the rise in Stage 3 can still be read mainly as a temporary congestion event. If recoveries stretch out, if a meaningful share remains on the watchlist in the next reports, or if Stage 2 continues to rise, the read changes. At that point it is no longer just a bridge-lending model that went through a busy year. It is a model where the middle layer is not clearing at the pace the growth strategy requires.

In a bond-only issuer, that is also how the market is likely to read the next phase. Not through another financing headline, but through three harder signs: does the watchlist contract, does Stage 2 move back instead of forward, and does the historical full-recovery record actually hold when the book is much larger.

Conclusion

The main article was right to mark the CMHC exit path as the heart of the story. This continuation shows where that heart is actually tested. By the end of 2025, the entire watchlist already sits outside Stage 1, and almost half of it is already in default. That is no longer a footnote on portfolio quality. It is the truth layer of the model.

The implication is that the 2026 question is not whether Peakhill knows how to roll loans. It is whether the watchlist can shrink before it becomes a larger Stage 3 problem. As long as that remains open, even a zero-realized-loss record is not enough to close the debate.

What has to be measured now: not only whether CAD 27.2 million is recovered, but whether the full CAD 56.3 million watchlist starts to come down, and whether CMHC exits begin to work faster than stage migration deteriorates.