A $390m construction loan says more about who takes the risk than about Walker & Dunlop

Walker & Dunlop, a commercial-property finance firm in Bethesda, Maryland, announced this week that it had "arranged" $390m of construction financing for Park Tower, a 47-storey, 1,049-unit tower rising in Jersey City's Journal Square, a five-minute walk from the PATH ride into Manhattan. Read the press release the way most investors do and it looks like evidence that the company is riding a revival in property dealings. It is evidence of that, but only obliquely. The word "arranged" carries nearly all the meaning, and it is the opposite of lending.

Walker & Dunlop did not put up the money. It brokered it. The financing is split into two floating-rate, interest-only loans — a senior construction loan provided by Affinius Capital and a mezzanine loan provided by BH3 Fund Advisors, two private-credit investors. Walker & Dunlop, acting as exclusive adviser to the developer Namdar Group, pocketed a fee for matching borrower with lenders and introducing the two capital providers to each other. The $390m sits on the lenders' books, not on the firm's.

That distinction matters because construction lending is among the riskiest corners of commercial-property finance. A building under construction earns nothing, so the loan pays no principal and the interest it pays floats with short-term rates. The only collateral is a half-built tower that must be finished, filled with tenants and sold or refinanced against whatever rents the market allows years later. The mezzanine piece, subordinate to the senior loan and first to absorb a loss, is the riskiest slice of the riskiest structure. Walker & Dunlop holds no part of it. It is the toll-collector on a road whose pavement, bridges and insurance belong to other people.

The toll road runs through shadow banking

The structure is nevertheless revealing, because it shows where that road now runs. Ten years ago a project like Park Tower would most likely have found its senior and mezzanine debt at a bank. No longer. Tighter regulation and heavier capital charges have pushed lenders out of development and transitional lending, and private-credit firms have moved in: they now account for an estimated 30–40% of non-agency commercial-property lending, up from almost nothing a decade ago. Walker & Dunlop's own numbers track the migration. Brokered loans made up 55% of its debt-financing volume in the first quarter, up from 49% a year earlier, and brokered lending grew 17% in the second quarter.

For the firm this migration is simultaneously the engine of its recovery and a thinning of its economics. The historic franchise is agency lending — originating apartment loans for Fannie Mae and Freddie Mac — which earns a fat origination fee and, because Walker & Dunlop keeps the servicing, a recurring fee stream that now runs across a $145.8bn servicing book. Brokered deals earn a smaller fee and generate no such stream. In the second quarter agency volumes fell 10% even as overall share rose, and management flagged that the tilt toward brokered, non-agency business was dragging on revenue. The machine that is growing is the one that pays least for each unit of volume.

A clean balance sheet, a scarred income statement

That tension shows up in the accounts. In the second quarter total transaction volume rose just 3% to $14.4bn, revenue fell 4%, and reported net income collapsed 91% to $3m. Most of that collapse, though, was not the transaction mix. It was a legacy hangover — $18m of credit costs tied to loans the firm was forced to repurchase after a fraud that Walker & Dunlop, working with Freddie Mac, attributes to a small group of sponsors. Strip out the noise and adjusted core earnings still rose 3%, carried by the unstoppable servicing book. The $390m headline is real news about a recovering pipeline of deals. It is not, by itself, an event for earnings.

Investors have already priced much of the ambivalence. The forward dividend yield stands near 6.75% — the market's way of saying it does not trust the economics to grow, or at least not yet, given how far the share has fallen. The question a beginner should pose is not whether one Jersey City tower gets built. It is whether the toll road is a good business to own the toll booth on. Handing risk to Affinius and BH3 keeps Walker & Dunlop's balance sheet clean, which is the point of its model and the source of its durability. But it is also the model's limit: a broker's fees are thin, and they shrink when the lenders it serves stop lending.

The deepest risk, then, is not on Walker & Dunlop's books at all. It is the pile of floating-rate, interest-only, mezzanine-laden construction debt that the firm has helped stack across gateway cities — debt that matures into whatever 2029's lease market looks like. If that wall of maturities cracks, the losses land on the private-credit investors, and the toll traffic thins. Walker & Dunlop has found a profitable, low-capital way to be paid for the migration of commercial-property risk from banks to shadow banks. Investors should read the happy headlines about it with the same sobriety they would apply to any middleman: the fee is real, the risk is not theirs, and neither, when the cycle turns, is the glory.