{"id":"094a94ba-79f6-4106-a0ae-bb3238ccc268","arxiv_id":"2508.14498","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Myopic profit-seeking firm reallocation with quadratic costs is a Wasserstein gradient flow whose long-run fixed point maximizes aggregate consumption, with several robustness extensions and EU firm-level calibration.","lead":"This paper models an economy where firms shift between sectors chasing higher profits and shows the process is mathematically a gradient flow that increases aggregate consumption over time. It also finds that EU sectoral profit rates converge while labor productivity does not, evidence for slow labor reallocation.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Externality-efficiency claim is asserted ('it can be shown') without proof; the advertised surprising result is unverified.","rationale":"I focused on the externality-efficiency claim because it is the most novel advertised contribution (abstract: 'surprising result that the decentralised equilibrium can remain efficient even in the presence of externalities') and because the supporting step is explicitly deferred ('it can be shown') without appearing in any appendix. The baseline model's gradient-flow and efficiency arguments are sketched and plausibly correct via Iacobelli et al.; the myopia/quadratic-cost assumptions are acknowledged limitations but not internal gaps. The missing proof is an omitted argument, not a matter of interpretation. The reader's weakest_assumption targeted behavioral premises; I partially agree, but the concrete missing proof is more decisive for the paper's central claim. A conditional verdict remains appropriate pending either a proof or an explicit restriction of the claim to the externality-free baseline.","tokens_in":38152,"tokens_out":10752,"duration_ms":124292,"concrete_test":"Compute the social-welfare functional C(ρ) (aggregate consumption) in the externality model from Proposition B.1. Derive the first-order condition for the constrained maximization of C(ρ) subject to ∫ρ=1, and compare it with (35). Check equality for all parameter triples satisfying (33). If ρ* does not maximize C, produce a two-sector counterexample. This directly verifies the 'surprising result'.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 2.5.1 (after Eq. (36)) states: 'it can be shown that ρ* maximizes F among all feasible distributions' and concludes that the decentralized equilibrium remains efficient even with positive externalities. This is a headline result of the abstract. No proof is given: Appendix B.1 only characterizes the short-run equilibrium with externalities; it never establishes that the functional F in (34) equals aggregate consumption C (or a monotone transform), nor that the stationary point (35) is the global maximizer of C. Condition (33) only guarantees well-posedness/gradient-flow existence. Since positive externalities normally create inefficiency, this claim needs a verifiable argument. The baseline Theorems 2.5-2.6 are proven, but the externality extension's efficiency conclusion is load-bearing for the paper's novelty and is currently unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a general-equilibrium model with a continuum of firms distributed over a circle of goods. At each instant a short-run competitive equilibrium determines sectoral profit rates; firms then reallocate myopically under quadratic adjustment costs, generating a PDE for the evolution of the firm distribution (Section 2.2, Eq. (21)). The main theoretical claim is that this PDE is a Wasserstein-2 gradient flow of a functional F, so that the decentralized dynamic can be viewed as a sequence of global optimization problems (Theorem 2.3 and JKO scheme), that aggregate consumption increases monotonically (Corollary 2.4), and that the unique long-run equilibrium maximizes aggregate consumption and equalizes profit rates (Theorems 2.5 and 2.6). The paper then extends the model to non-symmetric preferences, intrasectoral externalities, fixed reallocation costs, and immobile labour, and reports empirical estimates using ORBIS data on EU firms.","tokens_in":38239,"tokens_out":8333,"duration_ms":107896,"significance":"If the central claims hold, the paper is a substantial contribution: it gives a rigorous optimal-transport formulation of Sonnenschein–Artzner–Simon disequilibrium dynamics, provides a formal 'invisible hand' result for myopic profit-seeking with reallocation costs, and extends the representation to several economically meaningful frictions. The baseline theorems are nontrivial and appear internally consistent: the proof strategy via Iacobelli et al.'s weighted ultrafast diffusion class and the concavity argument for Theorem 2.6 are credible. The paper also ships reproducible code and data on Zenodo, and the empirical application to 680 EU sectors is a useful first pass. However, the novelty emphasized in the abstract and in Section 2.5.1—efficiency with positive externalities—is asserted rather than proved, and is thus load-bearing for the paper's claimed contribution.","major_comments":[{"comment":"The abstract's 'surprising result'—that the decentralized equilibrium remains efficient with positive (or mildly negative) externalities—is asserted with 'it can be shown' but no proof is supplied. Appendix B.1 gives the short-run equilibrium with externalities (Prop. B.1) and the fixed-cost example, but nowhere shows that the functional F in (34) is a monotone transform of aggregate consumption C in the externality case, nor that the stationary point (35) is the global maximizer of F or C. Condition (33) only ensures well-posedness. Since this claim is a headline contribution, please add a complete proof or explicitly weaken the claim.","section":"§2.5.1, after Eq. (36)"},{"comment":"The statements 'the rest of Theorem 2.3 still holds' and 'we can extend Theorem 2.5' to externalities are not backed by an argument in the appendix. The gradient-flow representation of Eq. (32), the JKO scheme, and global convergence to (35) require at least a proof sketch showing how condition (33) maps the equation into the Iacobelli et al. framework. Without this, the externality extension of the main theorems is only conjectural; this matters because the paper's abstract advertises the externality-efficiency result as one of its key findings.","section":"§2.5.1, before Eq. (35)"}],"minor_comments":[{"comment":"The inference that fixed costs are insignificant, based on the smoothness of the Nadaraya-Watson estimate in Figure 2, is not identified. A fixed cost creates an inaction region, but aggregation over five years and over 680 sectors could easily smooth this into a linear convergence pattern. Please either provide a formal test based on the implied plateau, or soften the abstract's claim of 'evidence suggesting the absence of significant fixed costs'.","section":"§4.1.1, Eq. (54)"},{"comment":"The fixed-cost equilibrium example uses a density that is unbounded at a point (it 'diverges at x=0'). The existence and uniqueness theorems in the paper assume bounded, strictly positive densities. Please clarify in what sense (39) is an admissible equilibrium, e.g., as a weak solution or after mollification.","section":"§2.5.2, Eq. (39)"},{"comment":"Several passages appear garbled, e.g., 'apart from' repeatedly appears where 'distribution' is presumably meant (Section 4.1.3), and the author affiliation line reads 'Giorgio F abbri'. Please proofread the final PDF carefully.","section":"Throughout"},{"comment":"The statement that 'the speed of convergence appears to be decreasing in ρ' is vague and not derived from Eq. (28). Either give the precise parametric dependence or remove this remark.","section":"§2.4, after Eq. (28)"}],"recommendation":"major_revision","confidential_remarks":"The baseline theory is sound and interesting, but the advertised externality-efficiency result is currently unsupported. If the authors can supply a proof in an appendix, the paper would be a strong contribution; if not, the claim should be removed from the abstract and Section 2.5.1 reframed accordingly. The empirical claims about fixed costs also need tempering or a sharper test."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: read the baseline model, don't take the externality-efficiency claim on faith. The baseline is a real contribution; the advertised surprising extension is unproven. What is new: the paper rewrites the multisector firm-reallocation PDE (myopic profit-seeking with quadratic costs) as a Wasserstein-2 gradient flow and gives a JKO scheme, so each step of the decentralized dynamics solves a global optimization of aggregate consumption net of reallocation cost. That is a genuinely different angle from Artzner-Simon-Sonnenschein's method of characteristics, and it produces the monotonicity and efficiency results in the baseline case. The proof route through Iacobelli et al.'s weighted ultrafast diffusion is sensible, and Theorem 2.6 is a clean concavity/Holder argument. The baseline results are not circular: the efficiency proof works against actual aggregate consumption. Soft spots, in proportion. The biggest one is Section 2.5.1. The claim that decentralized equilibrium remains efficient with positive intrasectoral externalities is stated as 'it can be shown' after Eq. (36), and no proof is given. The appendix doesn't establish that the functional F equals aggregate consumption (or a monotone transform) nor that the stationary point is the global maximizer of C. That matters because it is a headline result and because it contradicts standard externality intuition. As written, the result is unsupported. The fixed-cost extension is more honest, explicitly noting the loss of gradient flow and giving a constructed multiple-equilibrium example, but the general characterization is only heuristic. The empirical section is a light calibration: cross-sectional two-year/five-year changes, moderate R-squared, structural parameters recovered without confidence intervals. It supports the qualitative story but doesn't bear much weight. Credit where due: they ship code and data, cite the relevant literature, and are explicit about the myopia and quadratic-cost assumptions. Those assumptions are strong, but they are the model's premises, not hidden flaws. The paper would be stronger if it acknowledged the externality issue instead of burying the 'it can be shown.' A referee should be sent to verify the externality claim; it's likely either false or needs a non-obvious argument. Who gets value from this: theory economists working on disequilibrium dynamics or optimal transport methods in economics will get real value from Sections 2.3-2.6. The empirical part is skippable. This deserves peer review, but the revision needs to prove or soften the externality claim before I would trust the abstract's punchline.","headline":"The baseline gradient-flow representation is a genuine contribution; the headline externality-efficiency claim is asserted, not proven, so treat the paper as promising but needing revision.","tokens_in":735,"tokens_out":1575,"would_cite":true,"duration_ms":38226,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35K55","49Q22","91B50","91B62"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that uncoordinated, myopic profit-seeking firms collectively follow a path that maximizes aggregate consumption in the long run.","keywords":["gradient flow","Wasserstein space","emergent properties","myopic firms","sectoral reallocation","competitive equilibrium","invisible hand","labour immobility"],"falsifier":"Use firm-level panel data on sector switches to estimate the cost of moving from sector i to sector j as a function of economic distance between sectors. If the estimated cost is not approximately quadratic in distance, or if a fixed lumpy component dominates, the PDE is not the paper's gradient flow and the predicted monotone rise of aggregate consumption can fail. Alternatively, compute the paper's potential F from sector-level consumption data and test whether observed changes in sectoral profit rates follow its Wasserstein gradient; a rejection would falsify the central mechanism.","tokens_in":1737,"feed_emoji":"📈","tokens_out":2273,"duration_ms":76501,"temperature":0.7,"pith_summary":"This paper tries to show that Adam Smith's invisible hand is a precise emergent property rather than a metaphor. Firms move between sectors only to chase currently higher profit rates and pay quadratic costs for doing so; the economy as a whole then behaves as if it were solving a sequence of global optimization problems. The central result is that this uncoordinated, myopic reallocation makes a functional of aggregate consumption rise monotonically over time, and the economy converges to the unique long-run competitive equilibrium that maximizes aggregate consumption and equalizes profit rates across sectors. The proof works by recasting the partial differential equation governing the distribution of firms across a circle of goods as a gradient flow in Wasserstein space, so that the aggregate law of motion is the steepest-ascent path of an explicitly identified potential. The authors also show that fixed reallocation costs can break efficiency and uniqueness, that positive intrasectoral externalities can still leave the decentralized outcome efficient, and that immobile labour causes a measurable consumption loss. An empirical application to EU firms finds convergence of sectoral profit rates but not labour productivity, consistent with capital reallocating faster than labour.","feed_headline":"Myopic profit-chasing firms converge to an efficient equilibrium","feed_subtitle":"A gradient-flow proof shows uncoordinated profit-seeking drives the economy up a monotone consumption path.","key_machinery":"The central object is the Wasserstein-2 gradient flow on the space of firm distributions over a circle of goods, realized through the Jordan-Kinderlehrer-Otto (JKO) scheme. The space's distance is the minimal quadratic reallocation cost between two distributions, which matches the firms' adjustment cost structure. The potential F(μ), explicitly identified for the model's profit function, is log aggregate consumption, and the PDE of firm reallocation is the steepest-ascent dynamics of F in this Wasserstein geometry.","core_discovery":"The paper's central claim is that an economy in which firms move across sectors according to a myopic, profit-seeking rule with quadratic reallocation costs is exactly an optimizing system at the aggregate level. The density of firms over the circle of goods solves a continuity equation whose velocity is the local profit-rate gradient; Theorem 2.3 identifies this PDE as the Wasserstein-2 gradient flow of a potential F(μ) that is the natural logarithm of aggregate consumption. Consequently, along any trajectory F(μ) is nondecreasing (Corollary 2.4), and the firm distribution converges to the unique steady state that maximizes aggregate consumption and equalizes profit rates (Theorems 2.5 and","pith_inferences":["If reallocation costs scale with distance to a power other than two, the aggregate flow would live in a different Wasserstein space and the monotone quantity would change; the paper's proof does not cover this, but the JKO machinery suggests a testable family of models indexed by the cost exponent.","The result that myopic profit chasing remains efficient with positive externalities hints at a broader principle: whenever externalities are fully capitalized into current sectoral profit rates, decentralized reallocation can internalize them, and the paper's explicit example is one instance of that logic.","The paper leaves implicit a direct micro-level test of its key premise: reallocation flows should be proportional to the local profit-rate gradient, so firm-level data on sector switchers could be used to estimate the curvature of adjustment costs and thereby test the quadratic-cost assumption.","A natural extension, flagged by the authors, is joint capital and labour mobility with complementarities; that extension may produce agglomeration rather than equalization, reversing the paper's long-run efficiency conclusion."],"forward_implications":["If the central claim is correct, the sequence of short-run competitive equilibria is not arbitrary: it has a variational structure, with aggregate consumption serving as a Lyapunov function.","The baseline economy is globally stable and Pareto efficient in the long run, so disequilibrium dynamics self-organize to a welfare maximum without coordination or rational expectations.","With fixed reallocation costs, multiple long-run equilibria and persistent profit-rate dispersion can occur; the empirical evidence suggests such costs are small at the sectoral level.","With immobile labour, the consumption loss relative to the mobile-labour benchmark is minimized when labour is allocated as in the mobile economy, tying labour-mobility frictions directly to efficiency.","The empirical finding of profit-rate convergence alongside labour-productivity divergence is exactly what the model predicts when capital reallocates faster than labour."],"supporting_citations":[{"why":"Supplies the baseline mechanism of myopic firms moving along the goods space toward profit-rate differences and the method-of-characteristics convergence to long-run equilibrium.","marker":"Artzner et al. (1986)"},{"why":"Supplies the CES preferences and continuum-of-goods structure underlying demand, the price index, and the profit function.","marker":"Dixit and Stiglitz (1977)"},{"why":"Supplies the empirical grounding for quadratic capital reallocation costs, the key cost structure that makes the dynamics a Wasserstein-2 gradient flow.","marker":"Eisfeldt and Rampini (2006)"},{"why":"Supplies the JKO scheme, the variational approximation that lets the continuous-time dynamics be read as a sequence of global optimization problems.","marker":"Jordan et al. (1998)"},{"why":"Supplies well-posedness and the gradient-flow identification for the ultrafast diffusion equations used to identify the potential F and to prove convergence.","marker":"Iacobelli et al. (2019)"},{"why":"Supplies the theory of Wasserstein gradient flows, curves of maximal slope, and the monotonicity and convergence tools used in the core theorems.","marker":"Ambrosio et al. (2005)"},{"why":"Frames the economy as a sequence of short-run competitive equilibria with myopic firm adjustment, the conceptual structure the paper formalizes.","marker":"Sonnenschein (1982)"},{"why":"Supplies the cellular gradient-flow analogy linking single-unit rules to population-level dynamics and the social-utility monotonicity idea.","marker":"Horiguchi and Kobayashi (2023)"}],"fun_headline_variants":["Invisible hand emerges from myopic profit-seeking","Gradient flow makes invisible hand provably efficient","Myopic firm moves drive economy to efficiency","Uncoordinated profit-seeking optimizes aggregate consumption","Myopic reallocation converges to optimal allocation"],"cache_read_input_tokens":40576,"weakest_assumption_plain":"Firms' reallocation costs are exactly quadratic in distance and firms use today's profit rate as a complete guide to where to move; if either premise fails, the economy's path is no longer a Wasserstein-2 gradient flow and the monotone-consumption and efficiency theorems need not follow.","fun_headline_variants_meta":{"raw":{"variants":["Invisible hand emerges from myopic profit-seeking","Gradient flow makes invisible hand provably efficient","Myopic firm moves drive economy to efficiency","Uncoordinated profit-seeking optimizes aggregate consumption","Myopic reallocation converges to optimal allocation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000507,"raw_usage":{"total_tokens":2348,"prompt_tokens":825,"completion_tokens":1523,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":569,"completion_tokens_details":{"reasoning_tokens":1453}},"tokens_in":569,"tokens_out":1523,"duration_ms":14561,"temperature":1.0,"reasoning_tokens":1453,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T18:32:10.083567+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Use firm-level panel data on sector switches to estimate the cost of moving from sector i to sector j as a function of economic distance between sectors. If the estimated cost is not approximately quadratic in distance, or if a fixed lumpy component dominates, the PDE is not the paper's gradient flow and the predicted monotone rise of aggregate consumption can fail. Alternatively, compute the paper's potential F from sector-level consumption data and test whether observed changes in sectoral profit rates follow its Wasserstein gradient; a rejection would falsify the central mechanism.","supporting_citations":[{"cited_title":"P., and Sonnenschein, H","cited_arxiv_id":null,"evidence_quote":"Supplies the baseline mechanism of myopic firms moving along the goods space toward profit-rate differences and the method-of-characteristics convergence to long-run equilibrium."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the CES preferences and continuum-of-goods structure underlying demand, the price index, and the profit function."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the empirical grounding for quadratic capital reallocation costs, the key cost structure that makes the dynamics a Wasserstein-2 gradient flow."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the JKO scheme, the variational approximation that lets the continuous-time dynamics be read as a sequence of global optimization problems."},{"cited_title":"S., and Santambrogio, F","cited_arxiv_id":null,"evidence_quote":"Supplies well-posedness and the gradient-flow identification for the ultrafast diffusion equations used to identify the potential F and to prove convergence."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the theory of Wasserstein gradient flows, curves of maximal slope, and the monotonicity and convergence tools used in the core theorems."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Frames the economy as a sequence of short-run competitive equilibria with myopic firm adjustment, the conceptual structure the paper formalizes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the cellular gradient-flow analogy linking single-unit rules to population-level dynamics and the social-utility monotonicity idea."}],"review_version":1}