Policy and Welfare Analysis
Marketing research is usually asked whether something worked. This part is about the harder question that follows: whether it did any good, and for whom.
The distinction is not rhetorical. A rule that removes credit-card marketing from college campuses reduces card use, which is what it was designed to do; it also pushes students who still need liquidity into student loans, which nobody legislated and which no evaluation of the credit-card market can see (Brown, Grodzicki, and Medina 2026). An advertising ban reduces advertising in the banned category and moves demand to categories the ban did not reach (Dubois, Griffith, and O’Connell 2018). A privacy regime reduces tracking and reshapes which firms survive in the data supply chain (Johnson, Shriver, and Goldberg 2023). In each case the intended effect and the total effect are different objects, and the gap between them is where consumers actually end up.
These four chapters build the research architecture that closes the gap. The first treats spillovers as a formal estimand rather than a caveat, and shows how to find a credible control group for an economy-wide rule when the identifying variation has to come from differential exposure rather than differential policy. The second builds a normative benchmark—a small analytical model of what an informed, optimizing consumer would do—because the word “mistake” is meaningless without one, and because the model’s threshold rule is what tells the mechanism study which questions are diagnostic. The third derives a calibrated welfare bound, normalizing the unknown utility function away so that a signed conclusion survives an entire class of preferences rather than one convenient functional form (Bernheim and Rangel 2009; Chetty 2009). The fourth is about complementary evidence: a second test whose failure modes are disjoint from the first, so that agreement between them carries information rather than merely reflecting a shared assumption.
A fifth chapter turns the architecture toward publication. Almost every consequential consumer regulation of the past two decades restricts a marketing instrument—what may be advertised, to whom a firm may target, which data may be joined, which defaults may be pre-set, how a price may be displayed. Marketing owns the substitution structure of consumer choice and the measurement of instrument effectiveness, which is a comparative advantage the field has under-exploited in policy work. That chapter maps the architecture onto the marketing journals, sets out the sequence of moves these papers follow, and supplies a portfolio of project designs that are feasible with data that already exists.
The running case throughout is Title 3, Section 304 of the Credit Card Accountability, Responsibility, and Disclosure Act of 2009, and the study that traced its consequences into the student-loan market, through a matched survey of student decision-making, into a calibrated welfare bound, and finally into grade point averages and graduation rates. It is used not because consumer credit is the point but because it is the clearest available demonstration that a policy evaluation, a normative model, a welfare calculation, and a corroborating analysis can be assembled into a single argument in which each part answers the objection the previous part provokes.
The orientation of this part is the one the field’s editorial statements have asked for (Chandy et al. 2021; Wilkie and Moore 1999): research that takes consumer well-being as an outcome worth measuring, held to the identification and modeling standards of the quantitative track.