Description
Executive Summary:
-
Arbor Realty Trust is a REIT which provides short-term bridge financing for buyers of transitional multifamily housing
-
Arbor fell into the trap of ballooning its balance sheet with assets at the cycle peak (2021-mid 2022) before spiking interest rates and declining rent growth impaired the value of these projects.
-
We see Arbor common equity seriously impaired to wiped out by marking their book to economic reality, although before that outcome is realized Arbor is likely to cut and then eliminate its dividend, severely disappointing the retail investors who have flocked to the REIT.
-
The short-term nature of Arbor’s assets in tandem with their own highly leveraged inflexible balance sheet should force a reconciliation in relatively short order
Thesis:
Arbor Realty Trust is a commercial real estate mortgage REIT focused on making bridge loans to the multifamily property sector. We believe it is facing the type of existential challenges one might expect when a levered lender in credit and rate sensitive assets nearly triples its balance sheet over a 21 month period characterized by generationally low interest rates and unbounded enthusiasm over the prospects for the type of residential real estate that most benefited from stimulus. The prospects of dividend cuts, massively dilutive capital raises, both or worse might sound like hyperbole, but as we lay out below, our concerns are driven directly from the fundamental challenges Arbor faces and the simple math implied.
Bridge loans are short-term (2-3 years) loans at relatively high interest rates used to purchase and stabilize properties before hopefully refinancing them with long-term takeout debt. Arbor saw massive asset growth through the pandemic period when interest rates hit rock bottom levels and housing inflation skyrocketed. See https://www.propmodo.com/loans-supporting-troubled-multifamily-deals-nearing-maturity-causing-concern/. At YE2019 Arbor had roughly $7B of assets which have grown to approximately $17B today. The vast majority of the loans Arbor currently holds loans which were originated in the 2021-mid 2022 time period when the 10-year Treasury yield was between 0.50% and 1.5% versus 3.75% today.
A rating agency pre-sale report from one of Arbor’s 2021 bridge loan securitizations illustrates the aggressive assumptions underlying Arbor’s lending practices. Loans were made at 83% LTVs with the assumption that rent increases and capital investment could drive those LTVs towards 70% where the loans could be refinanced by the GSEs.
While Arbor does not provide detailed visibility into its portfolio, Trepp data shows the mean Arbor-financed property was built around 1967 and would be designated “Class C” multifamily. This is consistent with industry contacts which view Arbor as the C-class originator of multi-family loans to syndicators and small LLCs, relative to Walker and Donlop (WD) which is considered an A-class originator to institutionally sponsored assets.
Arbor has two big problems. First, the rapid rise in rates post-pandemic puts upward pressure on the debt burden of these projects and downward pressure on their valuations. Second, rent growth in their key Sunbelt markets has stalled,which makes it unlikely many of these projects will reach anywhere near their stabilized projections. Each of these problems challenge Arbor’s business model of getting these loans refinanced, but in concert they compound the issues because the gating factor to GSE takeout financing is debt service coverage ratio (DSCR). Fannie and Freddie require 1.25x including principal amortization at minimum. Higher interest costs and lower than projected rents pressure both the numerator and denominator of this ratio.
Let’s use an example to illustrate what this means. Assume that projects financed in the 2021/2022 time-frame were purchased at a 4.25% as-is cap rate. If the buyer paid $100 and borrowed at an 85% LTV there is $85 of bridge loan debt. The 4.25 cap rate implies the project throws off $4.25 in annual NOI. Current GSE multifamily rates run around a 200bps spread to the 10-year Treasury yield or approximately 5.75%. As mentioned, the most generous GSE financial programs require a 1.25x DSCR. Therefore, before even considering principal amortization, $4.25MM in NOI could support $59MM in debt at a 1.25x DSCR. But remember, Arbor extended $85 in bridge financing which is now $26 underwater or 30% of the loan’s principal balance.
But what about those rent increases and “stabilized” projections which would offset the rise in interest rates? Arbor finances its portfolio with a combination of CLO debt and repurchase facilities (repo) and for the half of its portfolio in CLOs we have relatively up to date transparency via Trepp into how things are going. We examined data through 1Q23 on 224 loans in Arbor CLOs with a total balance of $5.7B out of roughly $8B of Arbor CLO financed bridge loans. Those properties were generating $370MM in annual net cash flow. Again, using current GSE rates of 5.75% and a 1.25x minimum DSCR and again ignoring the fact that the GSE’s consider principal amortization when calculating DSCR,, $370MM would support maximum refinance proceeds of $4.1B which leaves a $1.5B hole for Arbor or 36% of the principal amount.
While these numbers are large they align with troubling qualitative evidence out of the multifamily sector. Nitya Capital appears to be Arbor’s single largest CLO financed borrower (again we don’t have transparency into the half of assets financed internally on repo lines) with $400MM of identified loans across nine properties. In February, Nitya sent its investors a letter admitting that multifamily properties it purchased in 2021 lost 44% of their value when interest rates rose.
SOURCE: Real Estate Alert
In a March 16 note, Barclays analyst Lea Overby commented that, “[Nitya] reportedly has $2bn of senior debt on its 60 properties. As short-term interest rates have risen, it is paying $60mn more annually in interest costs compared to what it paid in 2021. The article highlighted a Jacksonville portfolio which the firm bought in late 2021. While the rents on the portfolio have gone up by 22% under its ownership, the portfolio is cash-flow negative due to rising interest rate costs and the firm is covering the shortfall.”
Or consider the Houston Applesway portfolio which Arbor foreclosed on in the first quarter. According to The Real Deal, Arbor placed a credit bid on its own portfolio well below the principal amount due and won the bid as no other potential buyers emerged.These were 1970s-era Class C properties (which appears fairly representative of the overall Arbor portfolio). The property was ultimately sold for $196MM or a 13% discount to Arbor’s loan value. Did Arbor recognize the default and subsequent foreclosure as credit loss? No. Rather, Arbor provided 100% LTV financing to a new buyer to make this sale happen as the properties were in a tragic state of disrepair and in need of massive capital investment. In other words, the mid-teen percentage writeoff is probably understated as Arbor moved from what was allegedly an 85% LTV position to a 100% LTV position.
Rent growth is not coming to the rescue. Multifamily rents are no longer spiking as was seen in the 2021 period. Nationwide, effective asking rents on new apartment leases only increased by 0.3% as of April which is well below the rate of cost inflation for landlords (https://www.realtor.com/research/april-2023-rent/) and marked the 15th consecutive month of slowing rental growth. Even the sunbelt markets which saw the frothiest increases in housing prices are seeing weakness amidst demand destruction and near-record deliveries of new multifamily supply (https://www.costar.com/article/1372644014/sun-belt-no-longer-the-nations-multifamily-rent-growth-leader).
In fact multifamily rent growth is now running negative in some of the hottest housing markets of 2021 (where Arbor originations were concentrated) such as Jacksonville and Austin. Supply continues to come online in size putting downward pressure on renst with multifamily units under construction still at an all time high (See https://www.globest.com/2023/06/13/apartments-compete-for-new-renters-amid-burgeoning-supply/):
Can Arbor just extend and pretend on these loans until rent inflation bails them out? The answer to that is no as well. Arbor’s bridge loans along with their liabilities are variable rate and have a current yield around 8.5%. Just servicing the interest on $5.7B of borrowing would cost $485MM/year versus $370MM of property net cash flow. Much of Arbor’s borrower base purchased rate caps for protection against rising rates but they have run-off or will soon do so. Arbor holds $7.2B in multifamily bridge loans originated in 2021 and $4.8B originated in 2022. Arbor’s bridge loans mature in ~2.5 years at origination.As of 1Q the weighted average remaining months to maturity on their portfolio was 18 months. This means that the 2021 vintage loans are beginning to mature and will do so at scale over the course of 2023. As the corresponding interest rate caps expire many of these projects will not be able to maintain debt payments to Arbor in an extension scenario, much less meet strict DSCR requirements to obtain the requisite amount of takeout refinancing without significant external assistance. Furthermore, extensions would require Arbor to obtain new property value assessments to maintain compliance with its own contractual obligations to CLO bond holders and repo lenders. These appraisals as a rule will come in lower than where the lending facilities contemplate due to higher cap rates, lower rents and the mere fact that the need for an extension demonstrates the project did not meet its business plan.
Arbor itself is a highly leveraged borrower which constrains its ability to manage this problem. As of 1Q23 Arbor held $13.2B of bridge loans and $16.6B of total assets financed with $7.5B of securitized debt, $3.2B of repo facilities, $1.4B of senior unsecured notes, and $650MM of preferred stock for total leverage to the common equity of ~7x.
$7.5B of CLO debt was issued backed by a stated $8.5B in stated collateral value and $0.5B in restricted cash.
ABR retains the CLO equity tranches which are consolidated under GAAP accounting. CLO debt has the advantage of locked-in match funding but that assurance comes at a price.The most relevant covenant is the overcollateralization test. It requires the value of the assets in each CLO to maintain adequate cushion over the debt owed bondholders. By design, in order or maximize returns to the residual holder (Arbor) the overcollateralization is maintained very close to the minimum (see below) which effectively requires Arbor to purchase out of the CLO at par, any and all loans that run into trouble as failing to do so will cause the structures to immediately trap cash which means cash flow no longer flows to the residual holder. This is already happening and as described above, is very likely to occur at an accelerating and likely unsustainable pace.
Arbor also finances $4.4B of assets using $3.2B of repo financing. Few funding sources are more volatile than repo funding - especially for opaque level-three assets like bespoke bridge loans. Admittedly these are not overnight lines, but they do require annual renewal, with maturities staggered from mid 2023 through 2024. But as Arbor states in their 2022 10-K, “certain of our repurchase facilities include margin call provisions associated with changes in interest spreads which are designed to limit the lenders’ credit exposure. If we experience significant decreases in the value of the properties serving as collateral under these repurchase agreements, which is set by the lenders based on current market conditions, the lenders have the right to require us to repay all, or a portion, of the funds advanced, or provide additional collateral.”
If our analysis is correct, Arbor’s collateral is significantly impaired and if they can’t source sufficient liquidity to provide additional collateral they are at risk of catastrophic outcomes. All bank commercial real estate exposures must be getting a very high level of risk management and regulator attention right now and that likely means re-underwriting level 3 asset pricing assumptions. With lines coming up for renewal on a rapid basis (see Current Maturity column under Credit and Repurchase facilities exhibit above), Arbor’s repo lenders have multiple opportunities to increase haircuts (allow Arbor to borrow less on the same collateral), increase spreads (cost), lower limits, tighten collateral requirements or all of the above. Arbor's reliance on repo funding for illiquid assets is certainly a risk factor since these facilities are usually the first place trouble happens when wholesale financial markets break down due to panic or exogenous shock.
Arbor has a second related business originating multifamily loans for the GSEs. In fact, it is the balance sheet bridge loan business that feeds the GSE origination business. Arbor retains the servicing strip from originated loans and currently sits on a $29B servicing portfolio. Rising interest rates have been a nice boost for Arbor since it sits on considerable escrow funds. But this income also comes with considerable risk. At YE2022, 68% of Arbor’s servicing portfolio was with Fannie Mae where Arbor is part of their DUS (Delegated Underwriting and Servicing) program. DUS servicing fees are much higher than comparable GSE programs which is a nice tailwind in good times. But, as is described in Arbor’s 2022 10K this extra income is compensation for assuming significant credit and liquidity risk:
“Under the Fannie Mae DUS program, our Agency Business originates and services multifamily loans for Fannie Mae without needing Fannie Mae’s prior approval, as long as the loans meet the underwriting guidelines set forth by Fannie Mae. In return for such delegated authority and the commitment to purchase loans by Fannie Mae, we are required to share risk of loss on loans sold through Fannie Mae and we must provide collateral to Fannie Mae to secure any losses. Under the full risk-sharing formula, we absorb the first 5% of any losses on the UPB of a loan at the time of loss settlement, and above 5% we share the loss with Fannie Mae, with our maximum loss capped at 20% of the original UPB of a loan … At December 31, 2022, the Agency Business had pledged $64.4 million in restricted liquidity as collateral against future losses under $19.04 billion of loans outstanding that are subject to risk-sharing obligations. Fannie Mae collateral requirements may change in the future. At December 31, 2022, the Agency Business’s allowance for loss-sharing balance was $57.2 million, which may not be sufficient to cover future loss sharing obligations.”
…
For most loans serviced under the Fannie Mae DUS program, we are required to advance, in the event of a borrower failing to pay, the principal and interest payments and tax and insurance escrow amounts associated with a loan for four months. We are reimbursed by Fannie Mae for these advances, which may be used to offset any losses incurred under our risk-sharing obligations once the loan and the related loss share is settled.”
Independent of this credit risk, much of the volume from Arbor’s agency origination and servicing platform is generated by refinancing bridge loans from Arbor’s portfolio to the GSE as the terms of Arbor’s bridge financing make it prohibitive for a borrower to seek alternative financing. GSE multifamily originations are down generally but for Arbor specifically their shrinking portfolio eligible for refinance will weigh on origination fees.
Source: TREPP (https://www.trepp.com/hubfs/The%20Mid-Year%202023.pdf?hsCtaTracking=bd775462-8f7c-4142-86fb-561e22f954c8%7C09087725-16c8-4f3d-b98b-94c7ff7b1878)
Putting all of this together we see Arbor common equity as being wiped out, even in a base cae, under a mark to reality analysis. The short term nature of these loans in tandem with a levered inflexible balance sheet should make these problems evident sooner versus later:
It is always worth considering what scenarios could get Arbor out of its problems. First, if there is a macro outcome where 10-year rates decline considerably without enough associated economic weakness to pressure rents lower and spreads higher, they could squeeze their current portfolio through a refi window. Second, they could continue to benefit from the generosity of markets. Arbor is a prodigious issuer of equity via its ATM program. In the last two years Arbor raised $900MM in equity (~40% of its current book value). Arbor has a shareholder base dominated by index funds and retail dividend chasers. Even in 1Q23 when Arbor boasted of using its share repurchase program to support the stock, they issued $83MM of stock versus $9.7MM spent on the buyback program. Arbor is not the first company to realize that retail investors are not exactly great analysts of dividend sustainability. Eventually the dividend will have to be cut, but the longer Arbor can keep the ball in the air, the longer they can issue equity at what under this analysis is an extreme premium to underlying economic value.
After the market’s latest squeeze rally, Arbor is back to $14 which is approximately where it traded last summer when treasury rates were lower, slowing/declining rents were not yet evident, and catalysts driven by imminent maturities in 2021 vintage originations were still well in the future. As such we believe this is an opportune time to sell the stock.
Miscellaneous points
-
There is an existing short report on Arbor by Ningi Research (https://ningiresearch.com/2023/03/14/research-report-arbor-realty-trust-inc-nyse-abr/). I confess I did not follow most of it and can neither endorse nor refute its findings but I did find the section on Arbor’s escrow fees (“Enchanted Escrow Balances”) notable and perhaps worth a look for someone with better forensic accounting skills than my own. To be clear, we have no information or belief that management engaged in any actual improper behavior and want to focus our analysis and debate on pure math to the extent possible.
-
Shorting is hard and painful and timing can be difficult. Ivan Kauffman is a hardened survivor. We believe the math here is inevitable absent some unforeseen macro or regulatory deus ex machina. However, expect management to pull every lever to keep things afloat and play for optionality and recapitalization via share issuance. The retail investor base has a good chance of remaining stubbornly immune to contrary information. In other words, we hope you like the idea but caveat emptor.
-
We believe we are very conservative modeling 10% and 25% loan writedowns in the base/bear scenarios based on our bottom up valuation of the multifamily portfolio using Trepp securitization current NOI, our best assessment of cap rates for class c transitional multifamily properties, a qualitative assessment of Arbor’s publicly disclosed borrower quality, and the fact at Arbor could be a forced seller under certain liquidity scenarios. We admit to making some simplifying assumptions with the CLOs and examine them in aggregate rather than on a CLO by CLO basis but this does not materially change our directional conclusions.
v. 4.20
