Let me paraphrase what you’re asking, to make sure I understood. (Obviously if my paraphrase is wrong, then correct me.) I think you’re saying something like this:
Q: “I understand that someone might be willing to buy a one-year CD that yields only 2%, rather than a 5-year CD yielding an annualized return of 4%, because his money is only tied up for a year in the former asset. But OK, why wouldn’t someone buy the latter asset, hold it just for a year, and then sell it in order to reap the 4% return?”
A: I guess that’s right in the ERE (and if you think that means I said something wrong in my paper, please let me know), but it’s not right in a broader no-arbitrage concept of intertemporal equilibrium. The reason is that short-term interest rates might change in the future.
In particular, suppose you buy your 5-year CD, but that 6 months later interest rates across the yield curve all shift up by 50 percentage points. That means the market value of your CD drops; it is worth a lot less than you paid for it upfront. It’s true that if you hold it to maturity you still earn the contractual 4% annualized on your original investment, but you can’t even “get your money back out” of the thing until close to the maturity date.
So, given this possibility, if you really want to lock in a short-term return, you will keep rolling your money over in short-term CDs.
Does this make sense?