Popular Personal Financial Advice versus the Professors (James Choi, NBER)

I'm pretty sympathetic to the general idea of consumption smoothing over time based on one's expected average lifetime income; in particular, I think there are often cases where people are unnecessarily miserly with their money at a younger age in order to hit a certain saving rate, which isn't optimal for them.

However, one very important angle that informs a lot of this is the fact that we don't have efficient capital markets, which specifically here means that people don't have frictionless access to unlimited amounts of credit. This means that there's a very meaningful sense in which a "zero balance" is a meaningful lower bound, which means that it really does become important to save for emergencies, more than it otherwise would be (if your balance were allowed to go negative, then you wouldn't need to save for emergencies that much, because you could always finance emergencies with credit and pay it back with future earnings).

A related angle, that further amplifies this, is the significant uncertainty people may have about their life trajectory and future earning potential. Depending on risk-aversion, therefore, it may make sense to adopt the "consumption smoothing" strategy against not the median value of one's expected average income, but against a lower percentile (e.g., the 5th percentile or 25th percentile). In such situations, people would generally appear to see increasing consumption over time as it becomes clearer to them that they are not at the tail of worst outcomes with respect to lifetime income.

As an economist this makes me happy, but I feel it has to be pointed out that there are many assumptions underlying consumption smoothing that fail in consequential ways.

1. It assumes perfect credit markets. Saving is one thing, but if you think you'll earn more in the future, smoothing requires actually borrowing. And no bank will lend you money for personal consumption on the premise of "I'm going to be good for it in a few years, just trust me bro".
2. For many plausible utility functions, marginal utility is convex ($U^{\prime \prime \prime }\left(C\right)>0$) and the optimal thing to do is to save more when there is more uncertainty about your future income... which is exactly what most people do, rather than just focusing on their expected future income.
3. (Most importantly) For most of us, the importance of saving is not smoothing across time but smoothing across states. I save not for my 65 year old self, but for myself tomorrow who breaks a limb and has to pay a large healthcare payment. Smoothing consumption over states is literally impossible in a world where we can't draw perfect insurance contracts for every potential bad thing (or good thing) that we want to smooth over. When you have no insurance markets, it's pretty simple to establish that savings are the only way you have to smooth consumption across states.

So what should this mean for you? Well, a) your preferences matter; if you are "sufficiently risk averse" ($U^{\prime \prime \prime }\left(C\right)>0$) then save more than a risk neutral person would. b) saving is the only way to truly insure yourself across states, to guard against bad shocks to your livelihood and wellbeing. If you, like OP, want to do the "economically optimal" thing, I think you'd find that the answer will vary tremendously with your preferences and life circumstances, to the point where I think these frameworks should be taken as interesting ideas rather than prescriptions.

ARMs make sense in theory, but I've heard that in practice the embedded interest rate option tends to be underpriced because homeowners don't always exercise when it would benefit them.