Short-Term Optimal Recommendations and Long-Run Consumer Welfare: A Dynamic Model of Preference Evolution
Recommendation algorithms optimize click-through and conversion rates. Both are short-run objectives. As precision keeps rising, what happens to consumer welfare in the long run? We answer with a dynamic model in which preferences evolve endogenously through consumption experience. Goods fall into a mature category inside current tastes and an exploration category outside them, and even a forward-looking consumer can only choose within the recommendation set. Higher precision mechanically crowds out exploratory exposure. When the option value of exploration exceeds the current matching gain, long-run welfare decreases with precision and converges to a welfare lock-in level under full precision. Platform profit and consumer surplus then diverge systematically, which justifies a cap on precision. Numerical results show a welfare loss of up to 11 percent where exploration is valuable; the result flips when preferences are nearly exogenous. Regulation should target not the algorithm itself, but its compression of the exploration space.
Authors
- Jiakai Liu
Institutions
- University Town of Shenzhen (CN)
Publication Details
- Journal
- Economics & Business Management
- Published
- 2026-11-16
- DOI
- https://doi.org/10.63313/ebm.9240
- Primary Topic
- Digital Platforms and Economics
- Type
- article
- Field-Weighted Citation Impact
- 0.00