Work from home (WFH) has become a durable feature of the post-pandemic labor market, reorganizing spatial relationships among residences, workplaces, and neighborhoods. A large literature has documented the demand side of this shift—which jobs can be done at home, who adopts remote work, and how commuting and housing markets respond. Yet less attention has gone to the institutional layer that determines whether a neighborhood can legally absorb the resulting work activity: local land-use regulation, or zoning. Organized around the strict separation of home and work, Euclidean zoning may itself act as a friction that constrains local adaptation to an integrated live-work paradigm.

This paper asks how far local zoning codes enable or obstruct the spatial absorption of remote-work demand. When local regulation cannot accommodate this shift, excess demand may be capitalized into housing prices, expressed as income sorting, or realized as remote work below its structural potential. Measuring it has been difficult because zoning ordinances are fragmented across thousands of jurisdictions and written as unstructured legal text. But large language models now make them tractable at scale.

We assemble what is, to our knowledge, the largest national corpus of U.S. municipal zoning ordinances—8,225 jurisdictions, drawn from American Legal Publishing, Municode, and ordinance.com—and build a customized retrieval-augmented generation (RAG) pipeline that codes each jurisdiction's provisions directly from the legal text. The design grounds every coded value in a specific ordinance passage, preserving an auditable trail from measure to source. Validation confirms the measures' reliability: cross-model substitution (Qwen3.6, DeepSeek-V3.2, Claude Haiku 4.5) yields 92% majority agreement with the primary codes, and the extracted measures converge with external benchmarks (National Zoning Atlas, WRLURI).

Linking these measures to occupational WFH feasibility and 2019–2024 American Community Survey outcomes, we construct a WFH–zoning adaptation index that separates remote-work demand from local capacity to absorb it. Three findings emerge. First, pre-pandemic compositional structure—education, occupation, and industry mix—explains most cross-place WFH growth. Second, rules that directly govern home-based work are only weakly related to home-based activity, whereas indirect margins—accessory dwelling unit (ADU) and mixed-use allowances—predict both WFH growth and home-value capitalization, indicating genuine supply constraints rather than latent demand. Third, aligning demand against absorption identifies 528 reform-priority jurisdictions—concentrated in metros such as Boston, San Jose, and the Bay Area—where demand outstrips zoning capacity; by-right ADU reform alone closes roughly three-quarters of the gap. Together, these results trace a coherent mechanism: where zoning cannot absorb WFH demand, the shortfall surfaces as housing-market pressure and suppressed remote work, so the index locates not where demand is high but where institutional capacity binds.

The study advances the measurement literature along four dimensions. Empirically, it assembles the largest national corpus of municipal zoning text to date. Conceptually, it measures two margins existing indices have neglected: micro-level home-occupation allowances and neighborhood-scale commercial mixing. Methodologically, it demonstrates a transparent, replicable LLM pipeline for extracting theory-driven institutional measures from primary legal text at national scale. Practically, the resulting index supplies metro-specific evidence for zoning reform, pinpointing where remote-work demand meets local regulatory rigidity.