The "marriage premium" is the empirical observation that married people, on average, do better than unmarried people on a wide range of outcomes — income, wealth, health, longevity, mental well-being, child outcomes. The differences are not small. Married men earn more than unmarried men with otherwise similar profiles. Married couples accumulate substantially more wealth over a lifetime than otherwise similar singles. Married people live longer, get sick less often, and report higher life satisfaction. Children raised by their married biological parents fare better, on average, on educational, behavioral, and economic measures than children in other family structures. The premium shows up across cultures, across decades, and across most outcome measures researchers have looked at.
In 2000, Linda Waite and Maggie Gallagher synthesized this body of research into The Case for Marriage, arguing that marriage was not merely a tradition or a religious institution but a productive arrangement that conferred real and measurable benefits on its participants. The book was influential and became part of the policy case for marriage promotion in the United States, especially the Bush-era welfare initiatives that funded marriage-strengthening programs in low-income communities.
The critique came hard and from multiple directions. Philip Cohen and other family demographers argued that the premium was substantially — perhaps entirely — a selection effect. People who marry are systematically different from people who don't. They are healthier to begin with. They have more education, higher earning potential, more stable family backgrounds, fewer addiction issues, better mental health. These traits make them both more likely to marry and more likely to thrive. The marriage itself may add little; the people who walk into marriages were already on better trajectories. Comparing married to unmarried people therefore mostly compares two different populations, not two effects of marriage.
The selection critique is not the only one. Critics also pointed out that the premium varies enormously by who is being compared. The marriage premium for men is much larger than for women in income terms, partly because marriage frees men to specialize in earning while wives absorb domestic labor. The premium for women is contested and may be negative in pure income terms, though positive in wealth and health. The premium for children depends heavily on the quality of the marriage; children in high-conflict marriages fare worse than children whose parents separated. The premium for health is real but mediated by behavioral changes — married people exercise more, drink less, eat better — which means the proximate driver is behavior, not marriage per se.
There is also a temporal critique. The marriage premium has shrunk over time, especially as cohabitation has become a stable family form. The gap between married and cohabiting parents on most outcomes is smaller than the gap between either group and single parents. Much of what was once attributed to marriage is now attributable to having two committed, stably-coresident adults present, regardless of legal form. Marriage, in this reading, is one of several arrangements that produce the conditions for flourishing; it is not uniquely required.
The honest synthesis is that the marriage premium exists, that selection accounts for a meaningful share of it, and that what remains after selection is real but smaller than the headline numbers suggest. Marriage does seem to produce some genuine effects — through commitment, role specialization, mutual monitoring, shared resources, kin integration — that improve outcomes for participants. Those effects are not magical. They are the effects of stable, committed partnership combined with the social and legal scaffolding that marriage uniquely provides. Cohabiting unions can produce similar effects when they have similar stability, but cohabiting unions are on average less stable, which is why the premium persists.
The Law 2 task is to hold the complexity without collapsing it. The conservative case is correct that married households on average do better, and dismissing this as pure selection is unserious. The progressive case is correct that selection accounts for much of the gap and that policy aimed at "promoting marriage" without addressing the underlying conditions that make marriage hard for some populations is unlikely to work. Both sides of the debate have used the marriage premium to argue for their preferred policy. The premium is real, but the policy implications are not as direct as either side claims.
For the individual deciding whether to marry, the premium statistics offer thin guidance. Aggregate averages do not tell you whether this marriage to this person will produce the benefits. The premium is a population-scale phenomenon. Individual relationships are not. What the premium does tell us, collectively, is that stable committed partnership is a productive arrangement at scale — for participants and for the societies they make. Whether that arrangement is best instantiated as marriage, or as something else, is a question each generation will continue to answer.