Jue Theory Model

Spatial Theory & Model Craft (jue-theory-model)

When to trigger

• A reduced-form spatial result needs a mechanism that interprets the magnitude
• A referee asks "what does this estimate mean in equilibrium, once agents re-sort?"
• The paper wants counterfactuals (a policy, an infrastructure change) that require a quantitative spatial model
• The empirical object (a density-wage elasticity, a capitalization rate) maps to a structural parameter you have not named
• You are choosing how much model the paper needs: a one-equation Rosen–Roback wedge or a full QSM

How much model does a JUE paper need?

JUE is empirically led, but referees expect theory to discipline interpretation, not decorate it. Match the model to the claim:

Spatial-equilibrium discipline

1. Respect the indifference/zero-profit conditions. In a Rosen–Roback world a local change that looks like a pure benefit is partly capitalized into rents and partly offset by wage adjustment. A JUE referee will ask where the incidence falls — land, labor, or firms.
2. Close the model where agents move. If your empirical comparison holds location fixed but theory says agents re-sort, the reduced form is a short-run object; say so and bound the long-run.
3. For a QSM, tie every parameter to data. State which moment or reduced-form estimate identifies each elasticity (migration, commuting, housing supply, agglomeration). Report sensitivity of counterfactuals to the parameters that are least well identified.
4. Argue invariance for counterfactuals. The estimated elasticities must be policy-invariant enough for the experiment you run (a spatial Lucas critique).
5. Use theory to sign and bound, not to over-claim. The most persuasive JUE theory section delivers a comparative static the data then confirms.

Checklist

• The model is matched to the claim (capitalization / agglomeration / QSM / sorting / market access)
• Spatial-equilibrium conditions (indifference, zero-profit, market clearing) are respected
• Incidence is located: who bears the change — land, labor, or firms
• Short-run (location fixed) vs long-run (re-sorting) is distinguished
• QSM: each structural parameter is tied to an identifying moment/estimate; counterfactual sensitivity reported
• Counterfactual parameters argued policy-invariant
• Open-city vs closed-economy assumption stated and defended for the geographic scale
• The model is load-bearing (removing it makes an estimate uninterpretable), not decorative
• Theory yields a comparative static the empirics actually test

Anti-patterns

• A reduced-form result with no mechanism, leaving the magnitude uninterpretable
• Reading a capitalization estimate from prices alone, ignoring the wage and the worker indifference margin
• Attributing the density-wage elasticity to agglomeration when it is sorting of high-productivity workers
• A QSM whose counterfactual rests on parameters never tied to data, or whose elasticities are not policy-invariant
• Decorative theory: a model that does not change how any estimate is read
• Ignoring general-equilibrium reallocation, so a local gain is reported as a national welfare gain

Referee pushback mapped to the theory fix

• "What does this mean once households re-sort?" → Embed the estimate in a spatial-equilibrium model; report the short-run (location fixed) vs long-run (re-sorting) effect and where incidence lands.
• "Is this agglomeration or sorting of high types?" → Add the sharing/matching/learning micro-foundation and a decomposition that separates true spillovers from compositional sorting.
• "Your counterfactual parameters are not policy-invariant." → Argue invariance explicitly (spatial Lucas critique); show the elasticities are primitives, not functions of the policy.
• "The model is decorative." → Derive a comparative static the data then tests; if the model changes no estimate's interpretation, cut it.