6.8% unlevered on a lot I like, against demand I can't model

The demand question you can't model is partly answerable by asking the hospital. Do they lease spaces for staff, do they have a shuttle, are they building their own structured parking in the next five years? That last one would be the thing that hurts you and it's usually public in some form.

plumb's point about hospital-built parking is the real risk and I'd want it in writing somewhere before closing.

On structure, windlass: if you borrow, the covenants matter more than the rate at that yield. A DSCR test measured quarterly against parking income that swings with an event calendar and a weather month can trip on a bad quarter even when the asset is fine. Look at how income is defined, what cure rights you get, and whether there's a cash sweep. Also read any restriction on changing the use, because a lot of loan documents will limit you from doing exactly the redevelopment you're calling your floor. Confirm all of that in writing with the actual lender, since terms move and nobody here can tell you what's on offer this month.

Debt on cyclical revenue does one thing reliably, and I learned it the hard way over five years of running a service business: the payment shows up in the month your revenue didn't. Parking revenue is more variable at the transient end than people expect.

$64k net on 64 spaces is $1,000 a space a year. That's low for three blocks from a hospital if the rate is anywhere near market. What are you charging monthly, and are you full? If you're at 70% occupancy the answer to your debt question might be that you're buying a management problem at a 6.8% price and the fix changes the whole calculation.

cobalt is asking the right thing. From the lending mechanics, a lender on a 64-space surface lot is generally sizing to the dirt as much as to the income, because the improvements are asphalt and a kiosk. That means the loan may come in smaller than your income coverage suggests, and it may come from a local bank with a five year term and a recourse guarantee rather than anything long. So the practical answer to your question could be that heavy debt isn't on the menu regardless of what you'd prefer. Ask two local banks what they'd actually write on it before you decide what you want.

Moderate debt, and I'd size it so the payment is covered by the monthly contract base alone, ignoring all transient income. If the contracts cover debt service, the demand question you can't model becomes an equity return question instead of a survival question. That's how I've seen private capital handle assets with a soft long-term story.

I hit my own ceiling by buying at yields like this all cash and running out of money to buy the next thing. sable's coverage test is the version I'd use now. But windlass, you said the dirt is the floor and then didn't give us a land value. Three blocks from a hospital, what does the comparable unimproved parcel sell for, and is that number public where you are, because it isn't in every state? If the land is worth $700k of the $940k, this is mostly a land purchase with income attached and the debt question looks different than if the land is worth $300k.