The Strongest Evidence, Taken Seriously
Anchored to: Lusardi and Mitchell’s review and the Fernandes et al. meta-analysis, read together
Before I critique a theory I try to give it its best case, so here it is.
Lusardi and Mitchell (2014) reviewed the evidence and found that people with higher financial literacy tend to plan more and accumulate more wealth. That’s a real finding, and in the populations it was measured in, it holds up.
Now the harder number. Fernandes, Lynch, and Netemeyer (2014) pooled the research on financial education interventions and found they explained about one tenth of one percent of the variance in the behaviors studied. The effects faded with time, and they were weaker still in low-income samples.
Read together, these don’t contradict each other. They suggest that financial literacy matters a great deal where the conditions the theory assumes are present, and that teaching it does little where those conditions are absent.
That distinction is my own reading, and I’d welcome a challenge to it. But it explains something I couldn’t otherwise explain. Why would knowledge predict outcomes in one group and barely move them in another? Because knowledge only turns into behavior when the person can act on it. The conditions are part of the mechanism.
The claim I’m making is narrower than “education doesn’t work.” It’s that education is conditional, and the field has been treating the conditions as optional.
The practical upshot is a question I’d put to any program. Are the conditions the research assumed actually present for the people you’re serving? If they are, the strong literacy findings apply. If they aren’t, the weak intervention findings are the better guide, and the honest move is to change the conditions or the design, not to run the same course a third time with better slides.
I go through this evidence in this week’s episode.
If you design or evaluate programs, which of those two findings does your own data look more like?