First off, everybody's situation is different.
I used to scold the people who sell long dated boxes, and I preached selling short-dates ones, because with long-dated boxes you are basically selling the returns from the term premium to someone else. In other words, you are likely to pay more in the long run.
Also, consider that any longer-dated short box position is expected to be positively correlated to the stock market (because we expect long bonds to be negatively correlated to the stock market, especially during market crashes). So your overall portfolio risk probably increases, if your portfolio returns are dominated by the stock market i.e. if you have a large position in equities. It's kind of the opposite from your ability to get a regular loan during market crashes or recessions: With a margin account, you usually can get cheaper margin loans during bad times, i.e. your perceived "safety" from securing a longer-dated box is not "safety", but some additional risk.
EDIT: I think on the other hand there are times when the risk-free rate drops, but the repo rates briefly increase during really heavy market crashes.
I am now deliberating if I should revise my strategy. I personally use a leveraged equities + treasuries strategy with treasury futures, and boxes to leverage the equities side. Rolling treasury futures is clearly less labor-intensive than renewing SPX boxes at expiration. Therefore I am thinking of using longer-dated boxes, and "neutralizing" them i.e. neutralizing my duration exposure from the boxes by adjusting i.e. adding to the allocation to my treasury futures of about the same duration as my box accordingly.
I am trying to think this through. If we assume that the spread above the risk-free rate of boxes are constant with respect to the box maturities, then I think the only additional expense that I would incur is the expense from the implied spread of the additional treasury futures to compensate, right? Thoughts?