If you look at adults age 25-34 over the last twenty years, you see that the median net worth of married couples has held steady at around $25K per household (albeit with a temporary hit after the GFC) and that the median net worth of singles and unmarried couples has held steady at around $8K per household, all in real 2016 dollars. Overall, however, Americans in this age bracket have been losing ground. Why? Because the married share is shrinking. Back in 1987, 60% of them were married. Today, only 40% of them are. That, at least, is one way of cutting the data.

It is well known that--among all age brackets and in nearly all societies--married couples have always done a lot better economically than the unmarried (holding age, race, education, and other obvious variables constant). Social scientists have published reams of articles trying to untangle the causality behind this fact. Does affluence make marriage more probable (maybe because those with money or great career prospects find it easier to marry)? Of does marriage make affluence more probable (maybe by changing people's lifestyles in a way that favors career success)?

If forced to choose, most experts probably would favor yet another hypothesis: that an unobserved third factor is tipping the scales toward both marriage and affluence. A basic personality trait that is partly heritable and largely stable over a lifetime, for example, may be rigging the association. Consider likability and conscientiousness: People higher than average in these traits are both more likely to get married and to make more money. (Ditto for those low in neuroticism.) The evidence favoring this broader explanation is that married people turn out to have a lot of other unrelated outcomes in common: They are a lot happier; they get sick less; they live longer; they are more connected to the community; and so on.

Now let's get back to Millennials. For them, the third explanation does not help explain the rapid decline in the share who are getting married at all. But maybe the first explanation does. An undiminished share of Millennials want to get married. But Millennials have acquired--from older generations and from their own heightened perceptions of risk--certain standards of what kind of economic preparation is needed for a marriage to work. And now that fewer of them can meet that standard, fewer are getting married. And this may bring us back to the second explanation: If marriage itself is to some extent a causal generator of higher incomes and wealth accumulation, fewer Millennials are benefiting from that dynamic. (Unmarried cohabitation, though increasingly prevalent, does not seem to confer the same economic advantages.)

Sure, we'd like to know more about the extent to which marriage may actually cause affluence. Unfortunately, we cannot conduct randomized trials to test this causation--except, perhaps, on TV reality shows like A&E's "Married at First Sight."