Arbor Realty Trust (ABR) is a multifamily-bridge mREIT carrying $1B in non-performing assets through a second consecutive common-dividend cut ($0.30 → $0.17, -43%). Q1 2026 was reported May 8: common dropped 11.75% on the day; ABR-PE preferred dropped 0.47%. Three perpetual cumulative preferred series sit on top of the capital structure, and two of them trade at identical yields despite a structural asymmetry that should differentiate them.

What the filing shows

Q1 2026 distributable earnings were $0.18 against the new $0.17 dividend (94% payout). Management explicitly guided Q2 below cover at $0.15, then breakeven Q3, "growth" Q4. NPAs $1.002B (8% of $12.5B portfolio), down 10% QoQ as $274.8M of resolutions outpaced $197.8M of new delinquencies. PIK interest fell from ≈$18M/quarter year-ago to ≈$7M, suggesting genuine portfolio healing on the surviving book.

Preferred coverage is the load-bearing point for a preferred-share thesis. Q1 DE-before-preferred of $47.7M against quarterly preferred dividends of $10.3M = 4.6x earnings coverage. Common equity ≈$1.7B against total preferred liquidation preference ≈$657M = ≈22x asset coverage. The preferreds are cumulative — any suspended dividend accrues and must be paid back. ABR has paid every preferred dividend through this cycle, including both common dividend cuts.

CEO Kaufman opened prepared remarks declaring SEC/DOJ investigations "closed without any action against us" and the class action "dismissed without prejudice." The concurrent 8-K was Item 7.01 (investor presentation), not Item 8.01. No filing-grade confirmation of closure exists. In Q&A, analyst Rick Shane explicitly flagged a $25M REO sale paired with a $24.5M concurrent ABR bridge loan to the buyer (98% advance rate). A transcript-level scan across 11 CRE mREIT peers found no equivalent disclosed pattern. The binding test is Q2 10-Q related-party footnotes (late July).

The structural asymmetry

PF floats to ≈9.7% on October 29 at current SOFR. Refinancing PF saves $35-50M annually on $283M of paper. ABR has near-zero incentive to redeem PD or PE — both stay at fixed 6.25-6.375% indefinitely if held outstanding.

PF is the obvious refi target. PD and PE are both orphans relative to PF.

ABR-PD trades at $17.03 / 9.22% yield. ABR-PE trades at $16.91 / 9.20%. Yield differential: 2bps. Within the orphan pair, PE has the longer call window (94 days vs 28), the smaller float (5.75M vs 9.2M shares), and a more retail-skewed holder base — properties that should drive PE to widen relative to PD under issuer-stress, not trade in lockstep.

Why the gap might exist

Hypothesis: preferred desks default to treating sister issues from the same issuer as interchangeable when coupons and structures match. The call-priority sequence is implicit in prospectus terms across three separate documents and not synthesized in any sell-side note we located. ABR-PE has zero direct preferred coverage and ≈$97M float — too small for institutional pair-trade desks to size into. Market price-discovery on the spread depends on retail flow, which is driven by sentiment rather than capital-structure analysis. We can't prove no one has noticed; we can observe the spread is at zero.

Risks (ranked)

1. ABR redeems PD June 2 OR PE August 11. Removes the orphan-series asymmetry entirely. Probability 5-10%, contradicts management's stated capital-retention priority.
2. Class action refiled within 30-90 days. "Without prejudice" allows refiling with sharper pleading. Probability 35-45%. PE retail flight likely larger than PD.
3. Q2 10-Q discloses repeat REO-recycling transaction. Probability 50-60%. Asset-mark integrity concern intensifies, contagion across all three preferreds.
4. PF refi 8-K never lands. Probability 40-50%. Orphan logic stays unpriced indefinitely; PF floats with $35-50M/yr drag.

Catalysts

• 2026-06-02: ABR-PD first call (≈93% NOT redeemed)
• 2026-07-15 to 2026-08-08: Q2 2026 10-Q — highest-information disambiguator (REO footnote, dividend, NPA velocity)
• 2026-08-08: 90-day window for any 8-K confirming investigation closure (≈20-25% probability filed)
• 2026-08-11: ABR-PE first call (≈93% NOT redeemed)
• September-October 2026: ABR-PF refi 8-K window (≈50-60% probability)
• 2026-10-29: PF floats to ≈9.7% if not refinanced

What would change our mind

Confirms: ABR files 8-K announcing PF refi/redemption before October 12 — directly proves the call-priority sequence. PE/PD yield spread should compress and orphan-series logic gets recognized.

Kills: ABR redeems PD on June 2 or PE on August 11 (removes the asymmetry; capital deployed to redeem contradicts stated priorities). PE/PD yield spread widens beyond 100bps without a refi catalyst — implies hidden structural disadvantage in PE we're missing. Q2 10-Q discloses systematic REO-to-bridge recycling at scale that materially erodes book value — preferred liquidation cushion (≈22x today) collapses.

Evidence