Your 1.5% is almost certainly a portfolio average that already nets hedge benefit in, or a spot spread on new purchases. Those two numbers can differ by 60 basis points and the disclosure rarely makes it obvious which one is in the press release headline. Check the net interest spread table against the average balance schedule in the 10-Q and rebuild it yourself from interest income and interest expense.
The subtractions you're missing, in rough order of size. Management fee on externally managed vehicles runs around 1.5% of equity annually, and that's charged on equity, so at 7x it's a direct 150 basis points off your ROE, not off assets. G&A on top, often another 100 to 200 basis points of equity for a mid-sized manager. Hedge carry when the curve is unfriendly. And realized losses on portfolio turnover, which never appear in the spread line at all but very much appear in book value.
Premium amortization is the one that catches people. If you bought pools above par and prepayments run faster than the model assumed, you write off premium against interest income and the effective yield you actually earn is below the coupon. That mechanism alone explains a chunk of the drift between headline spread and realized ROE.
On the book value slide with a flat dividend: that combination usually means the dividend is being set off distributable earnings while mark to market losses run through equity. It's sustainable until it isn't. What I'd stress test is the repo book. Spread widening hits your assets and triggers margin calls at the same time, so the forced seller scenario is the one that turns a book value drift into a permanent impairment. Look at the maturity ladder on the repo and how much of it rolls inside 30 days.
