Staged Entry

Francesco Del Prato; Paolo Zacchia

REVIEW 2 major objections 4 minor 2 references

Staged Entry

T0 review · 2 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A model of two-stage firm entry shows that costlier, more precise gatekeeping can lower welfare even though the market's activation decision is constrained-efficient under CES preferences.

desk verdict A solid extension of Dhingra-Morrow to staged entry with noisy signals, whose main efficiency result appears correct but whose abstract overclaims a calibration and policy results that are not in the text. read the letter →

arxiv 2505.24460 v3 pith:V2MCZKYR submitted 2025-05-30 econ.GN q-fin.EC

Francesco Del Prato , Paolo Zacchia This is my paper

classification econ.GNq-fin.EC

keywords stagedentrygatekeepingmonopolisticcompetitionproductvarietywelfareCESpreferencesfirmselectionPigouviantaxation

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper models firm entry in two stages: entrepreneurs first pay a cost to learn a noisy signal of their productivity, then decide whether to pay a verification cost and activate, with true productivity revealed only after activation. It argues that under CES preferences the decentralized activation cutoff is constrained-efficient, because the consumer-surplus gain from an additional variety is exactly offset by business stealing. As a result, welfare losses from stricter gatekeeping do not reflect a private-social wedge and cannot be repaired by Pigouvian taxes or subsidies at the activation margin. Across regimes, higher screening precision improves selection but consumes resources and shrinks the mass of varieties; when the resource burden grows fast enough, welfare falls despite better average productivity.

What carries the argument

The key object is the two-stage entry structure with an activation cutoff $\theta^*$ and an operating productivity cutoff $\varphi^*$, pinned down by two equilibrium conditions: activation occurs when expected lifetime profit $\tilde{\pi}(\theta)/\delta$ equals the activation cost $f_b(\rho)$, and free entry equates the expected net payoff from experimentation to the experimentation cost $f_n$. The load-bearing identity is the CES proportionality between the planner's marginal welfare kernel and private surplus, $\tilde{G}(\theta)=\tilde{\pi}(\theta)/\delta - f_b(\rho)$, which makes the planner's activation rule identical to the decentralized rule. The welfare accounting then separates a selection term, $S(\rho)=E[\varphi^{\sigma-1} 1(\theta\ge\theta^*, \varphi\ge\varphi^*)]$, from resources per experimenter, $B(\rho)$, giving $d\log W/d\rho = (1/(\sigma-1))(S'/S - B'/B)$; welfare falls when the resource elasticity exceeds the selection elasticity.

What would settle it

Compute the constrained planner's marginal gain from activation without holding the operating productivity cutoff fixed; if a term involving the change in that cutoff survives in the planner's kernel, the market cutoff is not constrained-efficient. Alternatively, replace CES with a demand system with variable markups and check whether the planner's kernel is still proportional to $\tilde{\pi}(\theta)/\delta - f_b(\rho)$; any residual surplus term would overturn the no-Pigouvian conclusion.

Watch

Extended reading notes

Core claim

The central claim is that the welfare costs of costly gatekeeping are structural, not incentive-based. In the CES benchmark, the constrained planner's marginal payoff from activating an entrepreneur at signal $\theta$ is proportional to the entrant's expected profit net of the activation cost, $\tilde{\pi}(\theta)/\delta - f_b(\rho)$; the variety surplus the entrant cannot appropriate is exactly cancelled by the profit displaced from incumbents. The planner and the market therefore use the same activation cutoff, $\theta_P = \theta^*$, and a constant per-activation transfer with a compensating entry fee cannot raise welfare. What remains is a regime-level trade-off: higher verification precision raises average productivity through selection but acts like a higher fixed cost, shrinking experimentation and variety; under bounded cost schedules welfare can decline at high precision, and with unbounded costs welfare tends to zero near perfect precision.

Load-bearing premise

The load-bearing premise is that an extra activated firm's social value is exactly proportional to its private profit, with the consumer-surplus gain from a new variety just offsetting the business it steals from incumbents; if that offset fails, the market and planner activation cutoffs diverge.

Editorial extensions

If this is right

Tightening verification requirements can lower welfare even when the cost schedule is bounded and no private-social distortion exists.
Pigouvian taxes or subsidies timed to the activation decision are neutral within any fixed gatekeeping regime: welfare is maximized at a zero transfer.
Policies that raise compliance costs have first-order effects on firms that never apply, so gatekeeping should be evaluated on the experimentation margin, not only on survivors.
If verification costs grow without bound as precision approaches perfect information, the welfare-maximizing precision is interior and lies strictly below one.
Competitive financing of activation costs is allocation-equivalent to self-funding, so the results do not depend on credit frictions.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The exact offset behind the efficiency result relies on constant markups; under variable markups the consumer-surplus-from-variety and business-stealing terms need not cancel, so the no-Pigouvian conclusion is unlikely to survive outside CES.
A testable implication is a chilling effect on applications: if verification becomes costlier, experimenter entry should fall even among entrepreneurs who would have passed, and administrative data on application volumes by regulatory stringency could detect it.
The framework implies that gatekeeping welfare should be measured by the unseen extensive margin, the entrepreneurs who never apply, so comparing application rates before and after a compliance-cost change is a direct welfare-relevant experiment.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this is a well-constructed theory paper, and the central efficiency claim survives scrutiny, but the packaging overreaches. The model adds a two-stage entry process to the Melitz framework—entrepreneurs pay to learn a noisy signal, then decide whether to pay an activation cost—and shows that higher verification precision can reduce welfare. The constrained-efficiency result for the activation cutoff under CES (Lemma 1) is the load-bearing piece, and my own re-derivation confirms it: the general-equilibrium response of the operating cutoff cancels through the bivariate normal identities, leaving the planner's kernel proportional to private surplus. So the reader's worry about an omitted GE residual is real as a proof gap but not as a substantive flaw.

What is genuinely new: the staged-entry structure with imperfect, costly gatekeeping, and the welfare non-monotonicity result that ties losses to the resource cost of verification shrinking the feasible set of varieties rather than to a private-social wedge. The uniqueness proof in Appendix A.1 is careful, and the single-peaked FE locus argument is elegant.

Where the paper is soft: first, the abstract promises a calibration based on U.S. firm-entry data and Pigouvian policies for endogenous entry costs, but neither appears in the text. The only quantitative exercise is a parametric plot (Figure 3). That is a mismatch a referee would rightly flag. Second, Appendix A.7.1 is not a complete proof as written—it hides the planner's resource multiplier in an unspecified constant and does not explicitly track the operating cutoff—though, as noted, the result itself checks out. The authors should rewrite that appendix. Third, Proposition 3 is an existence result: for any two precision levels there exists a bounded cost schedule that makes welfare decline. That is honest but weaker than the abstract's blanket phrasing suggests. The Pigouvian irrelevance result is correctly stated as conditional on a fixed regime, but the conclusion in the main text occasionally reads as broader than that.

Who this is for: trade and IO theorists working on entry regulation, product variety, and the Melitz environment. It is a useful contribution even if it does not overturn a long-standing result; it extends one and adds a new margin. It deserves a serious referee. My recommendation: send it out, but require the authors to align the abstract with the content, expand A.7.1, and clarify the scope of Proposition 3.

Referee Report

2 major / 4 minor

Summary. The paper extends the closed-economy Melitz (2003) model to a two-stage entry process: entrepreneurs pay an experimentation cost to observe a noisy signal of productivity, then decide whether to pay an activation cost before true productivity is revealed. In the log-normal CES benchmark the paper characterizes equilibrium by a pair of cutoffs (θ*, φ*), proves uniqueness, and analyzes welfare. The central results are (i) Proposition 3, which shows that welfare can decline with verification precision even when verification costs are bounded, and (ii) Lemma 1, which claims that the decentralized activation cutoff is constrained-efficient conditional on the gatekeeping regime, so that welfare losses from stricter gatekeeping are a matter of the regime's resource cost rather than a private–social wedge. The paper concludes that no Pigouvian activation-margin policy can improve welfare.

Significance. If the efficiency result holds, the paper makes a useful conceptual point: under CES, the welfare cost of costly gatekeeping is a feasibility/variety effect, not a misallocation effect, and it connects cleanly to the Dhingra and Morrow (2019) optimality result. The welfare-nonmonotonicity result (Proposition 3) is a genuine and policy-relevant finding, and the log-normal benchmark provides closed-form expressions that allow sharp statements about equilibrium selection and welfare continuity. The paper also provides a competitive-intermediation microfoundation for the activation cost. The main weaknesses are that the proof of the key efficiency lemma is incomplete as written, and the abstract promises policy results and a calibration exercise that are not present in the body.

major comments (2)

[Appendix A.7.1] The derivation of the planner's kernel is incomplete. The argument holds the operating cutoff φ* fixed and absorbs the planner's resource multiplier into an unspecified positive constant κ, then concludes that the marginal social gain from activation has the same sign as π̃(θ)/δ - f_b(ρ). The suppressed general-equilibrium dependence of r̃(θ), π̃(θ), and the resource multiplier on the activation rule is precisely what Lemma 1, Corollary 2, and Corollary 3 hinge on. As written, the proportionality step does not show that the change in φ* induced by a marginal activation leaves the kernel's sign unchanged. A complete proof should either set up the constrained planner's problem with the resource constraint and derive the first-order condition in full, or show explicitly, using the bivariate-normal derivative identities, that the φ* response cancels. This is a proof gap rather than, in my assessment, a false result, but it must be fixed before the efficiency claim can be accepted as proven.
[Abstract and Section 1 (Contributions)] The abstract and introduction promise two things the manuscript does not deliver: an analysis of Pigouvian taxes or subsidies when entry costs depend on the number of firms passing an entry stage (knowledge spillovers or congestion), and a calibration exercise based on U.S. firm-entry data. The model in Section 2 takes f_b(ρ) as a function of precision only, with no dependence on the mass of entrants, and Figure 3 is an illustrative numerical example with chosen parameter values, not a calibration to U.S. data. These claims should be removed from the abstract and introduction, or the corresponding analysis and data work should be added. As it stands, the advertised scope of the paper exceeds its content.

minor comments (4)

[Section 2.2, Proposition 1] The AC locus is written as φ∗ = A(θ∗)ρ, which should read φ∗ = A(θ∗)^ρ (or φ∗ = A θ∗^ρ). The correct exponential form p∗ = ρ t∗ + a(ρ) appears in Appendix A.1.
[Appendix A.4 and A.5] The proofs contain the garbled token "/leftr⫯g⊸tl⫯ne/leftr⫯g⊸tl⫯ne→" in place of an arrow; this appears to be a rendering error and should be corrected.
[Section 2.1] The CES utility display is garbled: "U σ−1 σ = ∫ω∈Ω q(ω) σ−1 σ dω" should read U = (∫ q(ω)^{(σ-1)/σ} dω)^{σ/(σ-1)}.
[Section 1, Results] The phrase "a 'safer' market may be poorer" is evocative but not a formal claim; its formal content is in Proposition 3. Consider flagging it as an interpretation rather than a result.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Lemma 1 follows from a CES accounting identity with a common positive conversion factor, not from a fitted normalization, and the main welfare results are self-contained.

full rationale

Lemma 1 is not circular. Appendix A.7.1 computes the constrained planner's kernel from the CES accounting identity: welfare is the inverse price index, so the marginal welfare contribution of an activated entrepreneur is proportional to expected revenue r̃(θ)/δ; the induced labor cost is (r̃(θ)-π̃(θ))/δ + f_b(ρ); the net sign is r̃/δ - (r̃-π̃)/δ - f_b = π̃/δ - f_b. The constant κ is the common marginal-utility-of-income / resource-multiplier conversion; it multiplies both revenue and cost, so its positivity is all that matters and the conclusion does not depend on fitting it. This is an accounting identity, not a definition of the planner kernel as the private cutoff. The paper does not track the φ* response in A.7.1; that is an omitted envelope argument (the full-GE re-derivation confirms cancellation), not a circular reduction. There is no load-bearing self-citation: Dhingra and Morrow (2019) is external and provides independent support for the CES efficiency benchmark. The abstract announces a U.S. calibration and endogenous-cost Pigouvian policies that do not appear in the text; that is a missing-support/consistency defect, not circularity. Score 0.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The central claims rest on the CES monopolistic-competition structure inherited from Melitz/Dixit-Stiglitz, on Assumptions 1-3 (signal informativeness, costly gatekeeping, log-normality), and on the modeling choice that the constrained planner takes the mass of experimenters as given. No free parameters are estimated from data in the full text; the numerical illustration uses hand-picked values.

free parameters (8)

σ (elasticity of substitution) = 2
Set in the Figure 3 simulation; not estimated.
f (operating fixed cost) = 0.15
Hand-picked for the simulation in Figure 3.
f_n (experimentation cost) = 0.005
Hand-picked.
f_b0 (base activation cost) = 3
Hand-picked.
κ (cost curvature) = 2
Hand-picked.
α (precision exponent) = 8
Hand-picked.
δ (effective discount rate) = 0.1
Hand-picked.
log-normal marginal variances = 1 (normalized)
Assumption 3 sets standard normal marginals, fixing variance to 1; this is a normalization that affects quantitative results.

assumptions (5)

domain assumption Assumption 1: Signal informativeness (G(φ|θ) FOSD in θ)
Ensures cutoff behavior at the activation stage and monotonicity of expected profits.
domain assumption Assumption 2: Costly gatekeeping (f_b(ρ) weakly increasing in ρ)
Captures the resource cost of more precise verification.
domain assumption Assumption 3: Bivariate log-normal with standard normal marginals
Used for closed-form equilibrium characterization and the uniqueness proof.
standard math CES preferences and monopolistic competition
Inherited from Dixit-Stiglitz/Melitz; drives the markup constancy and the efficiency algebra.
ad hoc to paper Constrained planner takes the experimentation margin M_e as given
Defines the second-best benchmark; the first-best would also choose M_e.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Staged Entry." pith.science (2026). https://pith.science/paper/V2MCZKYR

@misc{pith2026250524460,
  author       = {Pith},
  title        = {Pith review of: Staged Entry},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/V2MCZKYR}},
  note         = {Machine review of arXiv:2505.24460}
}

read the original abstract

We develop a model of staged entry: to operate, monopolistically competitive firms must pay two sequential entry costs, each time acquiring a more informative signal of future performance. This model yields two implications for fiscal policy. First, the equilibrium outcome is constrained-efficient if preferences are CES and entry costs are exogenous. Second, when entry costs depend on how many firms pass any entry stage (due to positive knowledge spillovers or negative congestion effects) the resulting externalities can be offset via Pigouvian taxes or subsidies that are timed around the relevant entry decisions. A calibration exercise based on U.S. firm-entry data shows that these policies would raise welfare through a wider pool of entrants and sharper selection of productive firms.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

2 extracted references · 2 canonical work pages

[2]

Thus, write the FE condition as follows: H(p∗,t∗)=e 1 2 (σ−1)2−(σ−1)p∗ Φρ(−p∗+σ−1,−t ∗+ρ(σ−1) )− −Φ ρ(−p∗,−t∗)− δf b(ρ) f Φ(−t∗)− δf n f =0

√ 1−ρ 2 ⎞ ⎠ ϕ(t)dt −f ∫ ∞ t∗ Φ⎛ ⎝ ρt−p ∗ √ 1−ρ 2 ⎞ ⎠ ϕ(t)dt =f e 1 2 (σ−1)2−(σ−1)p∗ ∫ −t∗+ρ(σ−1) −∞ Φ⎛ ⎝ −ρh′−p ∗+(σ−1)√ 1−ρ 2 ⎞ ⎠ ϕ(h′) dh′ −f ∫ −t∗ −∞ Φ⎛ ⎝ −ρh−p ∗ √ 1−ρ 2 ⎞ ⎠ ϕ(h)dh =f[e 1 2 (σ−1)2−(σ−1)p∗ Φρ(−p∗+σ−1,−t ∗+ρ(σ−1) )−Φ ρ(−p∗,−t∗)], where the last line follows from the analysis of the standard normal cumulative distribu- tion’s moments as ...

work page 1980
[2016]

Liquidity Constrained Exporters

“Liquidity Constrained Exporters. ”Journal of Economic Dynamics and Control 72: 141–154. Dhingra, Swati, and John Morrow. 2019 . “Monopolistic Competition and Optimum Product Diversity under Firm Heterogeneity. ”Journal of Political Economy127 (1): 196–232. Dixit, Avinash K., and Joseph E. Stiglitz. 1977 . “Monopolistic Competition and Optimum Product Div...

work page 2019

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [2]

Thus, write the FE condition as follows: H(p∗,t∗)=e 1 2 (σ−1)2−(σ−1)p∗ Φρ(−p∗+σ−1,−t ∗+ρ(σ−1) )− −Φ ρ(−p∗,−t∗)− δf b(ρ) f Φ(−t∗)− δf n f =0

√ 1−ρ 2 ⎞ ⎠ ϕ(t)dt −f ∫ ∞ t∗ Φ⎛ ⎝ ρt−p ∗ √ 1−ρ 2 ⎞ ⎠ ϕ(t)dt =f e 1 2 (σ−1)2−(σ−1)p∗ ∫ −t∗+ρ(σ−1) −∞ Φ⎛ ⎝ −ρh′−p ∗+(σ−1)√ 1−ρ 2 ⎞ ⎠ ϕ(h′) dh′ −f ∫ −t∗ −∞ Φ⎛ ⎝ −ρh−p ∗ √ 1−ρ 2 ⎞ ⎠ ϕ(h)dh =f[e 1 2 (σ−1)2−(σ−1)p∗ Φρ(−p∗+σ−1,−t ∗+ρ(σ−1) )−Φ ρ(−p∗,−t∗)], where the last line follows from the analysis of the standard normal cumulative distribu- tion’s moments as ...

work page 1980

[2] [2016]

Liquidity Constrained Exporters

“Liquidity Constrained Exporters. ”Journal of Economic Dynamics and Control 72: 141–154. Dhingra, Swati, and John Morrow. 2019 . “Monopolistic Competition and Optimum Product Diversity under Firm Heterogeneity. ”Journal of Political Economy127 (1): 196–232. Dixit, Avinash K., and Joseph E. Stiglitz. 1977 . “Monopolistic Competition and Optimum Product Div...

work page 2019