So we're all familiar with the marshmallow test. I recently ran into a perfect real-life version. A former employer of mine actually has a defined benefit pension plan. I only qualified for it for a brief period of time, so I recently received the following offer:
I can either get a lump sum payment of nearly $8k today or get a monthly amount of ~$165 once I retire circa 2050. Obviously it's almost certainly better to take the DB payment since they're assuming all of the longevity and market risk and it diversifies my income stream in retirement, right? Not to mention the loss in taxes. So marshmallow test passed.
And yet, I've been thinking that maybe it doesn't make all that much sense to wait. The amount is a trifle (what will $165 buy each month in 2050? A few days of groceries?), and the hassle in keeping track of this information for the next 30-odd years is substantial. And I firmly expect to be capital rich but low consumption in early retirement while my current situation is decidedly the reverse; why not take a small boost to my buying power now over a largely irrelevant income stream in retirement?
I'm curious what you guys think - the amounts are essentially trifling so this decision has rather less import than most, but it made me think about the marshmallow test in a new light. It's traditionally seen as a way to determine the ability of a small child to delay gratification in anticipation of a greater reward. And yet this model ignores questions of time varying marginal utility: a single marshmallow now might indeed provide more overall utility than two marshmallows in the future. In fact, most NPV-type calculations also ignore this question; it's not necessarily a failing to value current consumption over future consumption, especially if the marginal marshmallow in the (marshmallow-rich) future is worth less than the marginal marshmallow in the (sadly, marshmallow-poor) present.