For now, I won't comment on the speculations about my motivation. In the future I will.
*The transaction could contain some kind of "flag" or other feature that would cause the transaction to no longer be valid according to consensus rules (due to a hard and/or soft fork) at the time the transaction becomes valid -- this is essentially what happened in July 2015 when a soft fork caused certain types of transactions to become invalid; someone signed and pushed a transaction (either because they were not aware the transaction was no longer valid, or
they signed it prior to the transaction becoming invalid), and the found block that confirmed that transaction was invalid and was forked from the main blockchain
*There could be a fork that causes any unspent bitcoin "in" addresses in their current form to be invalid either due to security issues with EDSCA or some other reason. Users would likely be given a lot of time to spend their bitcoin to a "new" address format, although with the NLockTransaction only being valid after a long time, it would be possible that the lender would not be able to get their bitcoin spent to a "new" address format.
*It is possible that the transaction will not contain a sufficient transaction fee to even propagate throughout the network (let alone confirm), and as a result the lender would never be able to actually spend their BTC
What about the risk of BTC/USD exchange price changing during longer terms, is that a risk/opportunity cost that concerns lenders or are we BTC permabulls here?
See above for most of my concerns. However for exchange rate concerns, it is possible that a lender would be made aware of some new information that might change their outlook on the price of bitcoin that would cause them to sell. Or it is possible that the USDBTC price could get to ridiculous levels that would cause the lender to wish to sell their bitcoin.
For the serious lender of 1000 BTC, I think the operational issues can be sorted out and so I quote above the concerns that I think are most relevant to our case, in addition to the potential regulatory intervention concern I mentioned in my prior post.
In particular the transaction fee issue is very worrisome given the block size debate and it is uncertain where transaction fees go in the distant future. But maybe we could simply assign this a cost, because I suppose it would be safe to assume that if we issue the transaction with a 5% fee then it will be okay. Also there could be several transactions prepared, each with different fees, but nevertheless the lender will need to access ROI based on the maximum fee that might become necessary in the future.
Combining the exchange rate risk, the regulatory intervention risk (e.g. what if they add MIT's ChainAnchor which requires KYC on all transactions), and your other two cited risks, it seems that for any long-dated nLockTime, the interest rate would need to be significant to compensate for risks.
Any guesses what level of interest rate the market might offer for this?
20% p.a.? 50%?