[mmcky]

Update the lecture to the October 2026 version of Alvarez and Argente, which adds terms in the aggregates alone and state-action cross terms, and split it into a core lecture and a sequel on extensions.

lq_mean_field_games (core):

• Give the planner the aggregate-only terms (2Θ + Λ); with them the capital-example planner adjusts faster than the equilibrium, and the Kimball planner's curvature is again free of the superelasticity
• Define the effective curvatures at the main result, move the individual/aggregate decomposition into the core theory, and put the four regions in the main text with a figure in the (θ_X, θ_A) plane
• Take the capital example's primitives out of the exercises, add a marginal-returns figure, and redo market power holding structural cost parameters fixed (η then drops out of aggregate persistence)
• Define the missing intercept, cite Wolf (2023), and replace a circular check with simulated panels
• State the positive-definiteness hypotheses behind the comparative statics, add the trace proposition and the paper's counterexample
• Make the Riccati solver reject non-solutions by checking the residual and the eigenvalue condition
• Reorder sections and exercises to follow the four takeaways, add a summary, and fix notation clashes

lq_mean_field_games_extensions (new):

• State-action cross terms, with a Hayashi adjustment-cost example and a one-state example of indeterminacy when the mixed interactions differ
• Forcing processes, with individual versus aggregate cost pass-through
• Coupling through higher moments

Also add the sequel to the table of contents and five bib entries.

[Copilot]

🔵 Needs a closer look

It is a large, mathematically dense two-lecture content change whose many quantitative economics results depend on notebook execution that cannot be verified in this environment, so it warrants expert human review.

No open issues 🎉

What changed in this PR

This PR revises the lq_mean_field_games lecture to track the October 2026 draft of Alvarez and Argente — adding aggregate-only return terms (so the planner's effective curvatures become 2Θ + Λ), defining "effective curvatures" at the main result, reorganizing the lecture around four takeaways, and hardening the Riccati solver — and splits the extension material into a new sequel lecture. Supporting changes wire the sequel into the table of contents and add five bibliography entries.

Changes:

• Rewrites lq_mean_field_games.md: effective-curvature framing, individual/aggregate decomposition, four-region figure, marginal-returns figure, market-power redone with structural cost parameters fixed, simulated identification panels, trace/eigenvalue propositions with a counterexample, and a solver that rejects non-solutions via residual + eigenvalue checks.
• Adds lq_mean_field_games_extensions.md: state-action cross terms (Hayashi example, indeterminacy when mixed interactions differ), forcing processes with pass-through, and coupling through higher moments.
• Registers the sequel in _toc.yml and adds Hayashi1982, CooperHaltiwanger2006, BenhabibFarmer1994, Wolf2023, BallRomer1990 to quant-econ.bib (plus type/month on AlvarezArgente2026).

File	Description
lectures/​lq_mean_field_games.md	Core-lecture rewrite: effective curvatures, four lessons, new figures, hardened mfg_riccati, planner with aggregate-only terms, reorganized exercises.
lectures/​lq_mean_field_games_extensions.md	New sequel covering cross terms, forcing processes/pass-through, and higher-moment coupling, with exercises.
lectures/​_toc.yml	Adds the sequel after lq_mean_field_games.
lectures/​_static/​quant-econ.bib	Adds five references and annotates the Alvarez–Argente working paper.

_{Review effort: Balanced}

[mmcky]

Review: looks good to merge once the preview build is green.

• Execution: both lectures run cleanly in a fresh kernel (core 42 cells in ~4 s, extensions 20 cells in ~2 s), using only NumPy/SciPy/Matplotlib.
• Math: re-derived the Hamiltonian structure, the solve_continuous_are mapping (including s = −C̃ for cross terms), the scalar formula and both thresholds, the planner terms (q + 2θ_X + Λ_X = ν(1−ν), Λ_A = (1/ε_s − 1)θ_A), the Hayashi algebra, the w-costate system, and the moment ODEs. No issues found.
• Prose vs output: every number checked matches the code output, and the two claims not computed in the lecture (firm half-life 2.3 → 3.4 across η; industry half-life 4.5 → 5.3 with c fixed) reproduce independently.
• Build hygiene: all cross-references resolve, no duplicate labels, the removed exercise labels aren't referenced elsewhere, and the five new bib keys are unique and cited.

I pushed 06b548b with style fixes only: five figure captions shortened to six words or fewer (qe-fig-004), and \mathbf y replaced by a plain y in the extensions (qe-math-004).

One optional suggestion: line 948 of the core lecture doesn't say how c is pinned down. The 4.5 → 5.3 numbers reproduce with c set so that γ = 0.05 at η = 4 and ε_s held fixed; a short parenthetical would let readers check it.

[mmcky]

thanks @thomassargent30

[github-actions]

📖 Netlify Preview Ready!

Preview URL: https://pr-1096--sunny-cactus-210e3e.netlify.app

Commit: 06b548b

📚 Changed Lectures

• lq_mean_field_games
• lq_mean_field_games_extensions

---

Build Info

• Workflow: Build Project [using jupyter-book]