Decision on SIG-2026-0163
0 views
Skip to first unread message
MSOM Conference
May 8, 2026, 4:26:04 PM (7 days ago) May 8
to d...@jhu.edu
08-May-2026
SIG-2026-0163, Government Intervention for Controlling African Swine Fever: Preemptive Biosecurity and Early Reporting
Decision: Major Revision
Dear Author (this is to ensure anonymity):
Thank you for submitting your manuscript to the M&SOM Healthcare Operations Management SIG. The paper was reviewed by an expert Associate Editor and two knowledgeable reviewers, and I am grateful to all three for the time, effort, and high-quality reports they provided. As you know, this paper was submitted under the SIG-Day + 1RR track: a Major Revision decision continues the paper at M&SOM with the same review team.
The paper addresses a consequential setting. The 2018 ASF outbreak in China destroyed roughly 40% of the country's herd, and the design of biosecurity-plus-indemnity regimes is a live regulatory question across the EU, East Asia, and the U.S. The modeling choice to separate a suspicion stage from disease confirmation is useful: it allows the substitution between biosecurity and reporting incentives to emerge from primitives rather than being assumed. Proposition 2's characterization of the optimal policy as a corner solution in the biosecurity-targeting space, with reporting compensation calibrated second, has a simple structure that could guide compensation design. Both reviewers acknowledge the importance of the problem.
At the same time, the review team has identified a set of concerns that the revision must address. Both reviewers recommend Major Revision, though R1 describes the decision as falling between Major Revision and reject-and-resubmit. Having read the paper and all three reports, I agree that the paper has a real contribution in it, but I want to be transparent that this is a high-risk revision. The concerns that follow are major, and the first one in particular is fundamental.
The single most important issue is the coherence of the information structure. R2 asks a carefully ordered set of questions: whether eta should act on S_i rather than D_i, whether the confirmation Bernoulli should really be independent of the other farm given that eta is a shock common to both, and whether D_i and D_j are truly independent. These are not clarification requests. They are asking whether the decomposition the paper adopts (eta acts on suspicion, confirmation is conditionally independent, and correlation propagates only through the suspicion layer) is the right one for the ASF context, or merely the tractable one. The AE fully agrees with R2 that this is the single most important issue for the revision, and so do I. An alternative decomposition in which eta acts on D_i with S_i as a noisy signal, or in which the common eta induces residual correlation in confirmation conditional on suspicion, would be equally natural and would change the comparative-statics logic. The authors need to (i) articulate why their chosen decomposition is the right one, grounded in ASF epidemiology, and (ii) verify which main results survive the alternatives. If this question cannot be resolved in a way that preserves the main results, the paper's core mechanism is at risk.
Second, several modeling assumptions need stronger defense or relaxation. The assumption that one farm's timely report fully protects the other farm is stated as reflecting real-world practice but is not derived from transmission dynamics; for ASF, which spreads through direct contact, fomites, and vectors, this is a strong assumption. The exclusion of penalty income from the government objective drives a result about higher penalties reducing welfare that needs to be stated and defended. And the binary biosecurity variable rules out interior effort levels, which is part of what delivers the corner-equilibrium result in Proposition 2. The AE and R1 both press these points.
Third, the paper needs sharper positioning against the existing literature. R1 asks for clearer articulation of what is new relative to Hennessy and Wolf (2018), Gramig et al. (2009) and Gramig and Horan (2011), Osseni et al. (2022), and related work. The AE suggests a framing along the lines of: "What becomes possible in our framework that is not in Hennessy and Wolf (2018)?" answered by a specific finding, with the asymmetric-reporting equilibrium under cost heterogeneity as the natural candidate. The AE also raises a useful point about the game-theoretic framing: while the Stackelberg label is descriptively correct, the paper's more distinctive contribution is about incentive design under private information, and connecting more explicitly to the principal-agent and mechanism-design literature on disease reporting would strengthen the positioning.
Fourth, the AE and R1 both note that the Theorem 2 result on trust heterogeneity follows rather directly from the modeling choice that trust enters as an observable discount on perceived compensation. If trust were private information, or affected reporting through additional channels, the policy conclusion could shift. The authors should discuss this limitation.
To summarize the priorities for the revision: (1) resolve the information-structure coherence question, which is fundamental; (2) defend or relax the key assumptions identified above; (3) sharpen the positioning against existing work and connect to the mechanism-design literature; and (4) discuss the limitations of the trust-heterogeneity result. The first priority is by far the most important, and I would urge the authors to begin there.
I want to reiterate that this is a high-risk revision. The information-structure question is not a matter of exposition; it goes to whether the model's core decomposition is the right one. If the authors can resolve it convincingly, the paper has a real contribution. I look forward to seeing the revised manuscript.
Thank you again for considering M&SOM.
Sincerely,
Tinglong Dai
MSOM Journal 1 RR Department Editor - Healthcare Operations
d...@jhu.edu
======================================================================
REVIEWER 1 REPORT
======================================================================
1. Please summarize the paper's contributions in a few sentences.
Please see the attached report.
2. What are the paper's strengths?
Please see the attached report.
3. Can you see this paper potentially being published in M&SOM, preferably in 2 rounds?
Please see the attached report.
Attached Referee Report:
Summary
The paper studies African swine fever (ASF) control in a Stackelberg game. The government chooses biosecurity effort and reporting-dependent compensation, and two farms decide whether to report suspected cases. The main findings are that (1) compensation increases reporting, (2) higher government biosecurity effort may reduce reporting by lowering farmers' perceived external risk, (3) the government can choose target levels of biosecurity effort and reporting, and then calibrate compensation to implement them. The paper also considers heterogeneity in reporting cost and trust, and includes an extension on on-farm biosecurity.
Overall assessment
My recommendation is between major revision and reject-and-resubmit. The core research topic is potentially interesting, but the current manuscript has important weaknesses in its positioning and relies on assumptions and modeling choices that are not sufficiently justified. The necessary revisions are substantial and may require nontrivial changes to the model with uncertain implications for the main results. For this reason, my recommendation is more negative than a typical major revision.
Major comments
1. The authors presents the inclusion of a suspected stage before confirmation, the two-farm structure, and the market-driven payoff as key contributions. These modeling choices may be useful, but the paper needs to articulate more clearly how they change the main insights relative to prior work. In particular, prior studies already examine reporting-related compensation, biosecurity effort, and the trade-off between them, and some of those findings appear largely consistent with this paper. Also, much of the analysis is based on two homogeneous farms. Therefore, the authors should clarify what new managerial insights are generated by these modeling features.
2. Many of the results seem to depend on strong assumptions. Several assumptions need stronger support from the literature or empirical evidence, rather than only verbal argument about intuition. The authors should also discuss more explicitly which results are robust under alternative settings.
a. The paper assumes that two farms face suspected cases and make reporting decisions at the same time, that the final pig price is market-driven, and that culling compensation is exogenous. In reality, suspected cases may occur at different times, and one farm's suspicion, report, or confirmation could affect the other's beliefs and behavior. The current game model can be a tractable approximation, but its implications should be discussed more carefully. Previous studies also show that compensation should be designed carefully to incentivize reporting. If pig price is market-driven, the treatment of culling compensation as exogenous needs stronger justification, especially given the prior literature emphasizing the importance of compensation design.
b. The authors assume that if a farm reports, the other farm is fully protected; if it does not report, both farms become infected and are culled. This strong assumption likely drives the clean results. Since prior epidemiological and economic studies often consider richer transmission dynamics, the paper should justify this assumption more carefully and discuss robustness under imperfect containment and probabilistic spread.
c. The paper assumes that the probability of a suspected case is lower when government biosecurity effort is higher. This is plausible for true infection risk, but it is less obvious for suspected cases, because suspected cases may include false alarms. The model also assumes that, conditional on one farm's suspicion, the probability that the other farm has no suspected case is nondecreasing in government biosecurity. These assumptions appear to be central to the result that stronger biosecurity may weaken reporting incentives, so they need stronger justification.
d. The authors assumes that suspicion is binary rather than continuous, excludes quick sale of unreported suspected pigs although previous literature identifies this as an important factor, and allows the two farms to differ mainly in reporting cost and trust while remaining identical in other dimensions. These simplifications are not necessarily inappropriate, but the paper should explain why they are reasonable and what may change if they do not hold.
e. The paper finds that higher penalties can reduce social welfare, but this result seems to rely on the modeling choice that government's objective does not include penalty income. The authors should clarify this assumption and explain its role in driving the result.
3. The trust heterogeneity analysis appears somewhat overly simplistic. In the current setup, trust affects only perceived compensation as a discount factor on the standard government payment. At the same time, the government is assumed to observe farmers' trust levels and to maximize welfare based on actual rather than perceived payoffs. Under these assumptions, it is not surprising that trust heterogeneity changes implementation cost but not the optimal target policy. The authors should discuss this result more carefully and explain how robust it is if trust is private information or affects reporting behavior more broadly.
Minor comments
1. Please check carefully whether the cited literature supports claims throughout the paper. For example, in the introduction, Guinat et al. (2016) do not directly support the claim that financial loss impacts reporting behavior, and EPRS (2023) does not have direct evidence for the compensation-related claims.
2. The authors should reconcile its discussion about government biosecurity efforts, reporting-related regulations, and testing. The paper states that government biosecurity includes surveillance and inspections, but also assumes that the government does not know about suspected cases unless the farmer reports it. Since immediate reporting of suspected cases is required in many regions, the paper should cite real examples of reporting violations and their consequences. The confirmation stage also needs more clarity, especially regarding what triggers testing if farmers do not report.
3. The paper should clearly present the functional forms of CS(), SE, and I(), and the corresponding objective functions.
4. The manuscript needs language editing. There are several typos and wording problems that reduce readability, such as "bioseucrity," "breading scale," and "descried."
Recommendation: Major Revision
======================================================================
REVIEWER 2 REPORT
======================================================================
1. Please summarize the paper's contributions in a few sentences.
The paper examines how governments should combine biosecurity measures and reporting incentives to control African swine fever. Using a game-theoretic model, it shows that compensation encourages reporting, while stronger biosecurity may reduce it. The optimal policy sets target levels for both and adjusts compensation accordingly, with flexibility needed depending on reporting costs and farm heterogeneity.
2. What are the paper's strengths?
This paper is well-written. I enjoyed reading it. It studies a practically relevant and interesting problem, and derives rich insights that are potentially useful to policymakers.
3. Can you see this paper potentially being published in M&SOM, preferably in 2 rounds?
I have several questions on the information structure of the model.
1. Bottom of page 8, the paper assumes that each farmer cannot observe the suspected status of the other farm, that is, farm i only knows its own S_i, and does not know S_j. In farming communities, farm owners may know each other well and share information with each other. First, under the model, do farms have incentives to share information? How are the results like under the alternative information structure where farm i knows both S_i and S_j?
I feel that my next three questions 2, 3 and 4 are related. But I couldn't wrap my head around to identify the root cause. I would like the authors to carefully reflect on/justify their assumptions. Currently, different modeling components do not appear to be coherent, with some even a bit contradictory.
2. Third paragraph of page 8: The government biosecurity effort \eta is modeled as influencing the probability that a suspected case occurs in a farm (i.e., S_i). Why is \eta not instead specified as affecting the probability of actual ASF (i.e., D_i)?
3. Second last paragraph of page 9: The paper assumes that, conditional on suspicion, the confirmation outcome can be modelled as an independent farm-level Bernoulli trial with success probability \theta, that is, P(D_i=0|S_i=1)=\theta. I am curious why this might not depend on farm j. In my mind, whether there is confirmed ASF in farm i (i.e., value of D_i) could be due to some common factors like the public health effort \eta that affects both farms, and some idiosyncratic factors specific to farm i.
4. In the analysis presented in Figure 2, it seems that the authors assume D_i and D_j (i.e., whether two farms have diseases or not) are independent. I find it a bit strange because the "signals" S_i and S_j are correlated because of the common public health effort \eta. Therefore, this common factor should also affect both the actual outcomes D_i and D_j.
My final question is related to the practical aspect of the insights.
5. It appears that the research question is broadly applicable to infectious diseases beyond ASF. The paper would benefit from a discussion section that expands on how the insights derived here could inform the control of other infectious diseases. In particular, it would be useful to consider whether different diseases are associated with varying magnitudes of reporting cost c or biosecurity effort cost I. If so, the authors could further enrich their insights by linking the main results in Figure 3 to a range of potential use cases.
Other minor questions
1. The authors may wish to be more precise in defining s_N and s_R (bottom of page 7). Both compensation schemes apply only when ASF is confirmed. In the current version, it initially appears that s_R is incurred whenever the farmer reports a case.
2. Prop 1: For the second line in Eq 8, are both probabilities between 0 and 1? Not clear from the expressions. Also, is the lower bound smaller than the upper bound?
3. What happens if there is heterogeneity in farm size?
Recommendation: Major Revision
======================================================================
ASSOCIATE EDITOR REPORT
======================================================================
1. Please summarize the paper's contributions in a few sentences.
This paper develops a Stackelberg game between a government and two pig farmers to study the joint design of biosecurity effort and reporting incentives for controlling African Swine Fever (ASF). The distinguishing modeling move is the explicit separation of a "suspicion" stage S_i from disease confirmation D_i: each farm privately observes whether it has a suspected case, updates its belief about the other farm via Bayes' rule, and then decides whether to report. Government biosecurity effort eta is modeled as acting on the suspicion probability p(eta, K_i); conditional on suspicion, confirmation is treated as an independent Bernoulli trial with P(D_i = 0 | S_i = 1) = theta. Three main findings follow: (i) reporting compensation s_R must clear a threshold to induce full reporting (Proposition 1); (ii) the optimal policy reduces to a choice among four corner equilibria in (eta, z_A, z_B) under homogeneous farms, with s_R flexibly calibrated after the targeting decision (Proposition 2); (iii) reporting-cost heterogeneity expands the policy space to six regions, letting the government target incentives toward low-cost farms while tolerating free-riding by high-cost farms (Proposition 3 / Theorem 1). Extensions treat trust heterogeneity and endogenous on-farm biosecurity.
2. What are the paper's strengths?
ASF is a consequential setting: the 2018 China outbreak alone destroyed 40% of the country's herd, and the design of indemnity-plus-biosecurity regimes is a live regulatory question in the EU, East Asia, and the U.S. The modeling choice to separate suspicion from confirmation is useful. It allows the biosecurity-reporting substitution (stronger biosecurity can unintentionally weaken reporting incentives by reducing perceived external risk) to emerge from primitives rather than being assumed. Proposition 2's characterization of the optimal policy as a corner in (eta, z_A, z_B) space with s_R calibrated second has a simple structure, and the welfare decomposition in Lemma 1 isolates the mechanisms clearly. The paper is generally well written, with careful notation (Table 1) and effective phase diagrams (Figures 3 and 4). Both reviewers, despite arriving at quite different overall assessments, acknowledge the paper addresses an important problem.
3. Can you see this paper potentially being published in M&SOM, preferably in 2 rounds?
The two reports. Reviewer 1 recommends Major Revision but describes the decision as falling between Major Revision and reject-and-resubmit; their concerns cluster around the sensitivity of the main results to modeling assumptions and the sharpness of the positioning relative to existing ASF literature. Reviewer 2 also recommends Major Revision and is enthusiastic about the problem and the insights, while pressing a carefully ordered set of coherence questions about the information structure. I am grateful to both reviewers for their time and careful engagement with the paper. These are serious reports, and the divergence in their overall assessments reflects a real tension in the paper rather than a difference in review quality. Having read the paper myself, I find that my own reading sits closer to the midpoint, and I want to organize my thinking around three issues.
1. Coherence of the information structure. Reviewer 2's questions about (a) whether eta should act on S_i rather than D_i, (b) whether the confirmation Bernoulli with P(D_i = 0 | S_i = 1) = theta should really be independent of farm j given that eta is a shock common to both farms, and (c) whether D_i and D_j in Figure 2 are truly independent: these are not clarification requests. They are asking whether the decomposition the paper adopts (eta acts on S_i; confirmation is conditionally independent; correlation propagates only through the suspicion layer) is the right one, or merely the tractable one. The paper's stance on p. 9 is that "correlation propagates from the suspicion stage," but this is asserted rather than derived. An alternative decomposition in which eta acts on D_i with S_i as a noisy signal of D_i, or in which the common eta induces residual correlation in D_i conditional on S_i, would be equally natural and would change the comparative-statics logic. To be clear, I do not think this means the paper is wrong, but the authors need to (i) articulate why the chosen decomposition is the right one in the ASF context, and (ii) verify which main results survive the alternatives. This is the single most important issue for a revision, and I fully agree with Reviewer 2 that the current manuscript does not yet answer it.
2. Assumptions that need stronger defense. Reviewer 1 identifies several assumptions I also found hard to accept without more grounding:
(a) Full protection if one farm reports. The payoff structure in Sections 3.1-3.2 (pp. 9-12) has the property that if farm i reports and D_i = 1, farm j is protected; if neither reports, both are culled after regional outbreak. This is stated as a reflection of "real-world regulatory practices" rather than derived from transmission dynamics. For ASF, which spreads by direct contact, fomites, and vectors across spatial proximity, this is a strong assumption, and it drives the payoff asymmetry that creates the reporting incentive. A brief epidemiological discussion, plus a robustness check under imperfect protection (e.g., j is protected with probability pi < 1), would go a long way.
(b) Penalty income excluded from the government objective. Reviewer 1 points out that the result that higher penalties can reduce social welfare relies on the government not internalizing penalty revenue. This is a reasonable modeling choice (penalties may be earmarked or administratively costly), but it needs to be stated and defended rather than left implicit.
(c) Binary eta and exogenous s_N. These are tractability-driven simplifications. The binary eta in particular rules out interior biosecurity levels, which is part of what delivers Proposition 2's corner equilibria. A brief discussion of what changes under continuous eta would help situate the contribution.
3. Positioning vs. the existing ASF / disease-reporting literature. Reviewer 1's first point is that the paper needs to articulate more sharply what is new relative to Hennessy & Wolf (2018), Gramig et al. (2009) and Gramig & Horan (2011), Osseni et al. (2022), and related work that already models reporting compensation, biosecurity, and their trade-off. The paper's declared innovations (the suspicion stage, the two-farm strategic structure, and the market-price payoff) are plausible candidates, but the paper does not work hard to show which of the main results actually depend on these innovations. My preferred framing would be: "What becomes possible in our framework that is not in Hennessy & Wolf (2018)?" answered by a specific finding. The asymmetric-reporting equilibrium under cost heterogeneity (Proposition 3 / Theorem 1) is a strong candidate, but the paper does not foreground it this way.
Game-theoretic framing. The paper's sequential structure, in which the government commits to biosecurity effort and reporting compensation and farmers respond, makes the Stackelberg label descriptively correct. But the more distinctive economic issue in the model is incentive design under private information: farmers privately observe suspicion and must be induced to report truthfully, while also interacting strategically with one another through reporting externalities and beliefs. For positioning purposes, the paper would benefit from connecting more explicitly to the principal-agent and mechanism-design literature on disease reporting and indemnity design. The concern is that on its own this framing understates the role of information asymmetry and implementation in the paper's contribution. The authors themselves adopt a mechanism-design framing in the Section 6 extension, which suggests they recognize this connection; making it explicit in the baseline model's positioning would strengthen the paper.
Trust heterogeneity (Theorem 2). I will briefly echo Reviewer 1 here. The Theorem 2 result, that trust heterogeneity changes the compensation level needed but not the targeting policy, follows directly from the modeling choice that trust enters as an observable discount delta_i on perceived compensation and the government maximizes true welfare. If trust were private information, or affected reporting through channels beyond delta_i * s_R (e.g., perceived detection probability delta_i * p_r), the policy conclusion could shift. The authors should discuss this.
The path question. This is a genuine split-recommendation case, and I want to be transparent about how I weigh it. On one hand, separating suspicion from confirmation is a real modeling move, and Proposition 2 has a simple structure that could guide compensation design. Reviewer 2's enthusiasm is earned. On the other hand, Reviewer 2's information-structure questions and Reviewer 1's concerns about assumption sensitivity and positioning are not cosmetic. A revision that (i) addresses Reviewer 2's questions 2-4 with both argument and robustness, (ii) defends or relaxes the assumptions in Section 3.1 and the payoff construction in Section 3.3, and (iii) sharpens the positioning against Hennessy & Wolf and the game-theoretic epidemiology literature would be a substantial program but is feasible in a single round, provided the authors agree that Reviewer 2's structural questions come first. If they cannot resolve (i) in a way that preserves the main results, the paper's core mechanism is at risk.
Recommendation: Major Revision
SIG-2026-0163, Government Intervention for Controlling African Swine Fever: Preemptive Biosecurity and Early Reporting
Decision: Major Revision
Dear Author (this is to ensure anonymity):
Thank you for submitting your manuscript to the M&SOM Healthcare Operations Management SIG. The paper was reviewed by an expert Associate Editor and two knowledgeable reviewers, and I am grateful to all three for the time, effort, and high-quality reports they provided. As you know, this paper was submitted under the SIG-Day + 1RR track: a Major Revision decision continues the paper at M&SOM with the same review team.
The paper addresses a consequential setting. The 2018 ASF outbreak in China destroyed roughly 40% of the country's herd, and the design of biosecurity-plus-indemnity regimes is a live regulatory question across the EU, East Asia, and the U.S. The modeling choice to separate a suspicion stage from disease confirmation is useful: it allows the substitution between biosecurity and reporting incentives to emerge from primitives rather than being assumed. Proposition 2's characterization of the optimal policy as a corner solution in the biosecurity-targeting space, with reporting compensation calibrated second, has a simple structure that could guide compensation design. Both reviewers acknowledge the importance of the problem.
At the same time, the review team has identified a set of concerns that the revision must address. Both reviewers recommend Major Revision, though R1 describes the decision as falling between Major Revision and reject-and-resubmit. Having read the paper and all three reports, I agree that the paper has a real contribution in it, but I want to be transparent that this is a high-risk revision. The concerns that follow are major, and the first one in particular is fundamental.
The single most important issue is the coherence of the information structure. R2 asks a carefully ordered set of questions: whether eta should act on S_i rather than D_i, whether the confirmation Bernoulli should really be independent of the other farm given that eta is a shock common to both, and whether D_i and D_j are truly independent. These are not clarification requests. They are asking whether the decomposition the paper adopts (eta acts on suspicion, confirmation is conditionally independent, and correlation propagates only through the suspicion layer) is the right one for the ASF context, or merely the tractable one. The AE fully agrees with R2 that this is the single most important issue for the revision, and so do I. An alternative decomposition in which eta acts on D_i with S_i as a noisy signal, or in which the common eta induces residual correlation in confirmation conditional on suspicion, would be equally natural and would change the comparative-statics logic. The authors need to (i) articulate why their chosen decomposition is the right one, grounded in ASF epidemiology, and (ii) verify which main results survive the alternatives. If this question cannot be resolved in a way that preserves the main results, the paper's core mechanism is at risk.
Second, several modeling assumptions need stronger defense or relaxation. The assumption that one farm's timely report fully protects the other farm is stated as reflecting real-world practice but is not derived from transmission dynamics; for ASF, which spreads through direct contact, fomites, and vectors, this is a strong assumption. The exclusion of penalty income from the government objective drives a result about higher penalties reducing welfare that needs to be stated and defended. And the binary biosecurity variable rules out interior effort levels, which is part of what delivers the corner-equilibrium result in Proposition 2. The AE and R1 both press these points.
Third, the paper needs sharper positioning against the existing literature. R1 asks for clearer articulation of what is new relative to Hennessy and Wolf (2018), Gramig et al. (2009) and Gramig and Horan (2011), Osseni et al. (2022), and related work. The AE suggests a framing along the lines of: "What becomes possible in our framework that is not in Hennessy and Wolf (2018)?" answered by a specific finding, with the asymmetric-reporting equilibrium under cost heterogeneity as the natural candidate. The AE also raises a useful point about the game-theoretic framing: while the Stackelberg label is descriptively correct, the paper's more distinctive contribution is about incentive design under private information, and connecting more explicitly to the principal-agent and mechanism-design literature on disease reporting would strengthen the positioning.
Fourth, the AE and R1 both note that the Theorem 2 result on trust heterogeneity follows rather directly from the modeling choice that trust enters as an observable discount on perceived compensation. If trust were private information, or affected reporting through additional channels, the policy conclusion could shift. The authors should discuss this limitation.
To summarize the priorities for the revision: (1) resolve the information-structure coherence question, which is fundamental; (2) defend or relax the key assumptions identified above; (3) sharpen the positioning against existing work and connect to the mechanism-design literature; and (4) discuss the limitations of the trust-heterogeneity result. The first priority is by far the most important, and I would urge the authors to begin there.
I want to reiterate that this is a high-risk revision. The information-structure question is not a matter of exposition; it goes to whether the model's core decomposition is the right one. If the authors can resolve it convincingly, the paper has a real contribution. I look forward to seeing the revised manuscript.
Thank you again for considering M&SOM.
Sincerely,
Tinglong Dai
MSOM Journal 1 RR Department Editor - Healthcare Operations
d...@jhu.edu
======================================================================
REVIEWER 1 REPORT
======================================================================
1. Please summarize the paper's contributions in a few sentences.
Please see the attached report.
2. What are the paper's strengths?
Please see the attached report.
3. Can you see this paper potentially being published in M&SOM, preferably in 2 rounds?
Please see the attached report.
Attached Referee Report:
Summary
The paper studies African swine fever (ASF) control in a Stackelberg game. The government chooses biosecurity effort and reporting-dependent compensation, and two farms decide whether to report suspected cases. The main findings are that (1) compensation increases reporting, (2) higher government biosecurity effort may reduce reporting by lowering farmers' perceived external risk, (3) the government can choose target levels of biosecurity effort and reporting, and then calibrate compensation to implement them. The paper also considers heterogeneity in reporting cost and trust, and includes an extension on on-farm biosecurity.
Overall assessment
My recommendation is between major revision and reject-and-resubmit. The core research topic is potentially interesting, but the current manuscript has important weaknesses in its positioning and relies on assumptions and modeling choices that are not sufficiently justified. The necessary revisions are substantial and may require nontrivial changes to the model with uncertain implications for the main results. For this reason, my recommendation is more negative than a typical major revision.
Major comments
1. The authors presents the inclusion of a suspected stage before confirmation, the two-farm structure, and the market-driven payoff as key contributions. These modeling choices may be useful, but the paper needs to articulate more clearly how they change the main insights relative to prior work. In particular, prior studies already examine reporting-related compensation, biosecurity effort, and the trade-off between them, and some of those findings appear largely consistent with this paper. Also, much of the analysis is based on two homogeneous farms. Therefore, the authors should clarify what new managerial insights are generated by these modeling features.
2. Many of the results seem to depend on strong assumptions. Several assumptions need stronger support from the literature or empirical evidence, rather than only verbal argument about intuition. The authors should also discuss more explicitly which results are robust under alternative settings.
a. The paper assumes that two farms face suspected cases and make reporting decisions at the same time, that the final pig price is market-driven, and that culling compensation is exogenous. In reality, suspected cases may occur at different times, and one farm's suspicion, report, or confirmation could affect the other's beliefs and behavior. The current game model can be a tractable approximation, but its implications should be discussed more carefully. Previous studies also show that compensation should be designed carefully to incentivize reporting. If pig price is market-driven, the treatment of culling compensation as exogenous needs stronger justification, especially given the prior literature emphasizing the importance of compensation design.
b. The authors assume that if a farm reports, the other farm is fully protected; if it does not report, both farms become infected and are culled. This strong assumption likely drives the clean results. Since prior epidemiological and economic studies often consider richer transmission dynamics, the paper should justify this assumption more carefully and discuss robustness under imperfect containment and probabilistic spread.
c. The paper assumes that the probability of a suspected case is lower when government biosecurity effort is higher. This is plausible for true infection risk, but it is less obvious for suspected cases, because suspected cases may include false alarms. The model also assumes that, conditional on one farm's suspicion, the probability that the other farm has no suspected case is nondecreasing in government biosecurity. These assumptions appear to be central to the result that stronger biosecurity may weaken reporting incentives, so they need stronger justification.
d. The authors assumes that suspicion is binary rather than continuous, excludes quick sale of unreported suspected pigs although previous literature identifies this as an important factor, and allows the two farms to differ mainly in reporting cost and trust while remaining identical in other dimensions. These simplifications are not necessarily inappropriate, but the paper should explain why they are reasonable and what may change if they do not hold.
e. The paper finds that higher penalties can reduce social welfare, but this result seems to rely on the modeling choice that government's objective does not include penalty income. The authors should clarify this assumption and explain its role in driving the result.
3. The trust heterogeneity analysis appears somewhat overly simplistic. In the current setup, trust affects only perceived compensation as a discount factor on the standard government payment. At the same time, the government is assumed to observe farmers' trust levels and to maximize welfare based on actual rather than perceived payoffs. Under these assumptions, it is not surprising that trust heterogeneity changes implementation cost but not the optimal target policy. The authors should discuss this result more carefully and explain how robust it is if trust is private information or affects reporting behavior more broadly.
Minor comments
1. Please check carefully whether the cited literature supports claims throughout the paper. For example, in the introduction, Guinat et al. (2016) do not directly support the claim that financial loss impacts reporting behavior, and EPRS (2023) does not have direct evidence for the compensation-related claims.
2. The authors should reconcile its discussion about government biosecurity efforts, reporting-related regulations, and testing. The paper states that government biosecurity includes surveillance and inspections, but also assumes that the government does not know about suspected cases unless the farmer reports it. Since immediate reporting of suspected cases is required in many regions, the paper should cite real examples of reporting violations and their consequences. The confirmation stage also needs more clarity, especially regarding what triggers testing if farmers do not report.
3. The paper should clearly present the functional forms of CS(), SE, and I(), and the corresponding objective functions.
4. The manuscript needs language editing. There are several typos and wording problems that reduce readability, such as "bioseucrity," "breading scale," and "descried."
Recommendation: Major Revision
======================================================================
REVIEWER 2 REPORT
======================================================================
1. Please summarize the paper's contributions in a few sentences.
The paper examines how governments should combine biosecurity measures and reporting incentives to control African swine fever. Using a game-theoretic model, it shows that compensation encourages reporting, while stronger biosecurity may reduce it. The optimal policy sets target levels for both and adjusts compensation accordingly, with flexibility needed depending on reporting costs and farm heterogeneity.
2. What are the paper's strengths?
This paper is well-written. I enjoyed reading it. It studies a practically relevant and interesting problem, and derives rich insights that are potentially useful to policymakers.
3. Can you see this paper potentially being published in M&SOM, preferably in 2 rounds?
I have several questions on the information structure of the model.
1. Bottom of page 8, the paper assumes that each farmer cannot observe the suspected status of the other farm, that is, farm i only knows its own S_i, and does not know S_j. In farming communities, farm owners may know each other well and share information with each other. First, under the model, do farms have incentives to share information? How are the results like under the alternative information structure where farm i knows both S_i and S_j?
I feel that my next three questions 2, 3 and 4 are related. But I couldn't wrap my head around to identify the root cause. I would like the authors to carefully reflect on/justify their assumptions. Currently, different modeling components do not appear to be coherent, with some even a bit contradictory.
2. Third paragraph of page 8: The government biosecurity effort \eta is modeled as influencing the probability that a suspected case occurs in a farm (i.e., S_i). Why is \eta not instead specified as affecting the probability of actual ASF (i.e., D_i)?
3. Second last paragraph of page 9: The paper assumes that, conditional on suspicion, the confirmation outcome can be modelled as an independent farm-level Bernoulli trial with success probability \theta, that is, P(D_i=0|S_i=1)=\theta. I am curious why this might not depend on farm j. In my mind, whether there is confirmed ASF in farm i (i.e., value of D_i) could be due to some common factors like the public health effort \eta that affects both farms, and some idiosyncratic factors specific to farm i.
4. In the analysis presented in Figure 2, it seems that the authors assume D_i and D_j (i.e., whether two farms have diseases or not) are independent. I find it a bit strange because the "signals" S_i and S_j are correlated because of the common public health effort \eta. Therefore, this common factor should also affect both the actual outcomes D_i and D_j.
My final question is related to the practical aspect of the insights.
5. It appears that the research question is broadly applicable to infectious diseases beyond ASF. The paper would benefit from a discussion section that expands on how the insights derived here could inform the control of other infectious diseases. In particular, it would be useful to consider whether different diseases are associated with varying magnitudes of reporting cost c or biosecurity effort cost I. If so, the authors could further enrich their insights by linking the main results in Figure 3 to a range of potential use cases.
Other minor questions
1. The authors may wish to be more precise in defining s_N and s_R (bottom of page 7). Both compensation schemes apply only when ASF is confirmed. In the current version, it initially appears that s_R is incurred whenever the farmer reports a case.
2. Prop 1: For the second line in Eq 8, are both probabilities between 0 and 1? Not clear from the expressions. Also, is the lower bound smaller than the upper bound?
3. What happens if there is heterogeneity in farm size?
Recommendation: Major Revision
======================================================================
ASSOCIATE EDITOR REPORT
======================================================================
1. Please summarize the paper's contributions in a few sentences.
This paper develops a Stackelberg game between a government and two pig farmers to study the joint design of biosecurity effort and reporting incentives for controlling African Swine Fever (ASF). The distinguishing modeling move is the explicit separation of a "suspicion" stage S_i from disease confirmation D_i: each farm privately observes whether it has a suspected case, updates its belief about the other farm via Bayes' rule, and then decides whether to report. Government biosecurity effort eta is modeled as acting on the suspicion probability p(eta, K_i); conditional on suspicion, confirmation is treated as an independent Bernoulli trial with P(D_i = 0 | S_i = 1) = theta. Three main findings follow: (i) reporting compensation s_R must clear a threshold to induce full reporting (Proposition 1); (ii) the optimal policy reduces to a choice among four corner equilibria in (eta, z_A, z_B) under homogeneous farms, with s_R flexibly calibrated after the targeting decision (Proposition 2); (iii) reporting-cost heterogeneity expands the policy space to six regions, letting the government target incentives toward low-cost farms while tolerating free-riding by high-cost farms (Proposition 3 / Theorem 1). Extensions treat trust heterogeneity and endogenous on-farm biosecurity.
2. What are the paper's strengths?
ASF is a consequential setting: the 2018 China outbreak alone destroyed 40% of the country's herd, and the design of indemnity-plus-biosecurity regimes is a live regulatory question in the EU, East Asia, and the U.S. The modeling choice to separate suspicion from confirmation is useful. It allows the biosecurity-reporting substitution (stronger biosecurity can unintentionally weaken reporting incentives by reducing perceived external risk) to emerge from primitives rather than being assumed. Proposition 2's characterization of the optimal policy as a corner in (eta, z_A, z_B) space with s_R calibrated second has a simple structure, and the welfare decomposition in Lemma 1 isolates the mechanisms clearly. The paper is generally well written, with careful notation (Table 1) and effective phase diagrams (Figures 3 and 4). Both reviewers, despite arriving at quite different overall assessments, acknowledge the paper addresses an important problem.
3. Can you see this paper potentially being published in M&SOM, preferably in 2 rounds?
The two reports. Reviewer 1 recommends Major Revision but describes the decision as falling between Major Revision and reject-and-resubmit; their concerns cluster around the sensitivity of the main results to modeling assumptions and the sharpness of the positioning relative to existing ASF literature. Reviewer 2 also recommends Major Revision and is enthusiastic about the problem and the insights, while pressing a carefully ordered set of coherence questions about the information structure. I am grateful to both reviewers for their time and careful engagement with the paper. These are serious reports, and the divergence in their overall assessments reflects a real tension in the paper rather than a difference in review quality. Having read the paper myself, I find that my own reading sits closer to the midpoint, and I want to organize my thinking around three issues.
1. Coherence of the information structure. Reviewer 2's questions about (a) whether eta should act on S_i rather than D_i, (b) whether the confirmation Bernoulli with P(D_i = 0 | S_i = 1) = theta should really be independent of farm j given that eta is a shock common to both farms, and (c) whether D_i and D_j in Figure 2 are truly independent: these are not clarification requests. They are asking whether the decomposition the paper adopts (eta acts on S_i; confirmation is conditionally independent; correlation propagates only through the suspicion layer) is the right one, or merely the tractable one. The paper's stance on p. 9 is that "correlation propagates from the suspicion stage," but this is asserted rather than derived. An alternative decomposition in which eta acts on D_i with S_i as a noisy signal of D_i, or in which the common eta induces residual correlation in D_i conditional on S_i, would be equally natural and would change the comparative-statics logic. To be clear, I do not think this means the paper is wrong, but the authors need to (i) articulate why the chosen decomposition is the right one in the ASF context, and (ii) verify which main results survive the alternatives. This is the single most important issue for a revision, and I fully agree with Reviewer 2 that the current manuscript does not yet answer it.
2. Assumptions that need stronger defense. Reviewer 1 identifies several assumptions I also found hard to accept without more grounding:
(a) Full protection if one farm reports. The payoff structure in Sections 3.1-3.2 (pp. 9-12) has the property that if farm i reports and D_i = 1, farm j is protected; if neither reports, both are culled after regional outbreak. This is stated as a reflection of "real-world regulatory practices" rather than derived from transmission dynamics. For ASF, which spreads by direct contact, fomites, and vectors across spatial proximity, this is a strong assumption, and it drives the payoff asymmetry that creates the reporting incentive. A brief epidemiological discussion, plus a robustness check under imperfect protection (e.g., j is protected with probability pi < 1), would go a long way.
(b) Penalty income excluded from the government objective. Reviewer 1 points out that the result that higher penalties can reduce social welfare relies on the government not internalizing penalty revenue. This is a reasonable modeling choice (penalties may be earmarked or administratively costly), but it needs to be stated and defended rather than left implicit.
(c) Binary eta and exogenous s_N. These are tractability-driven simplifications. The binary eta in particular rules out interior biosecurity levels, which is part of what delivers Proposition 2's corner equilibria. A brief discussion of what changes under continuous eta would help situate the contribution.
3. Positioning vs. the existing ASF / disease-reporting literature. Reviewer 1's first point is that the paper needs to articulate more sharply what is new relative to Hennessy & Wolf (2018), Gramig et al. (2009) and Gramig & Horan (2011), Osseni et al. (2022), and related work that already models reporting compensation, biosecurity, and their trade-off. The paper's declared innovations (the suspicion stage, the two-farm strategic structure, and the market-price payoff) are plausible candidates, but the paper does not work hard to show which of the main results actually depend on these innovations. My preferred framing would be: "What becomes possible in our framework that is not in Hennessy & Wolf (2018)?" answered by a specific finding. The asymmetric-reporting equilibrium under cost heterogeneity (Proposition 3 / Theorem 1) is a strong candidate, but the paper does not foreground it this way.
Game-theoretic framing. The paper's sequential structure, in which the government commits to biosecurity effort and reporting compensation and farmers respond, makes the Stackelberg label descriptively correct. But the more distinctive economic issue in the model is incentive design under private information: farmers privately observe suspicion and must be induced to report truthfully, while also interacting strategically with one another through reporting externalities and beliefs. For positioning purposes, the paper would benefit from connecting more explicitly to the principal-agent and mechanism-design literature on disease reporting and indemnity design. The concern is that on its own this framing understates the role of information asymmetry and implementation in the paper's contribution. The authors themselves adopt a mechanism-design framing in the Section 6 extension, which suggests they recognize this connection; making it explicit in the baseline model's positioning would strengthen the paper.
Trust heterogeneity (Theorem 2). I will briefly echo Reviewer 1 here. The Theorem 2 result, that trust heterogeneity changes the compensation level needed but not the targeting policy, follows directly from the modeling choice that trust enters as an observable discount delta_i on perceived compensation and the government maximizes true welfare. If trust were private information, or affected reporting through channels beyond delta_i * s_R (e.g., perceived detection probability delta_i * p_r), the policy conclusion could shift. The authors should discuss this.
The path question. This is a genuine split-recommendation case, and I want to be transparent about how I weigh it. On one hand, separating suspicion from confirmation is a real modeling move, and Proposition 2 has a simple structure that could guide compensation design. Reviewer 2's enthusiasm is earned. On the other hand, Reviewer 2's information-structure questions and Reviewer 1's concerns about assumption sensitivity and positioning are not cosmetic. A revision that (i) addresses Reviewer 2's questions 2-4 with both argument and robustness, (ii) defends or relaxes the assumptions in Section 3.1 and the payoff construction in Section 3.3, and (iii) sharpens the positioning against Hennessy & Wolf and the game-theoretic epidemiology literature would be a substantial program but is feasible in a single round, provided the authors agree that Reviewer 2's structural questions come first. If they cannot resolve (i) in a way that preserves the main results, the paper's core mechanism is at risk.
Recommendation: Major Revision
MSOM Conference
May 8, 2026, 5:18:58 PM (7 days ago) May 8
to msom-confe...@googlegroups.com
08-May-2026
Re: SIG-2026-0163, "Government Intervention for Controlling African Swine Fever: Preemptive Biosecurity and Early Reporting"
SIG Day Decision: Reject
Dear Author (this is to ensure anonymity):
We received many excellent submissions for the Healthcare Operations Management SIG-Day Conference. Unfortunately, we were unable to accept all of them to be included in the program, and we are sorry to say that your paper was not accepted to the SIG-Day conference.
If you also submitted an extended abstract of your paper to the main MSOM Conference, a decision on that submission will come separately.
Sincerely,
Healthcare Operations;SIG Co-Chairs
---------------------
Referee: 1
Please summarize the paper briefly. - Rev: Please see the attached report.
Referee: 2
Please summarize the paper briefly. - Rev: The paper examines how governments should combine biosecurity measures and reporting incentives to control African swine fever. Using a game-theoretic model, it shows that compensation encourages reporting, while stronger biosecurity may reduce it. The optimal policy sets target levels for both and adjusts compensation accordingly, with flexibility needed depending on reporting costs and farm heterogeneity.-
Referee: 1
Comments to the Author (old)
(There are no comments.)
Referee: 2
Comments to the Author (old)
(There are no comments.)-
Re: SIG-2026-0163, "Government Intervention for Controlling African Swine Fever: Preemptive Biosecurity and Early Reporting"
SIG Day Decision: Reject
Dear Author (this is to ensure anonymity):
If you also submitted an extended abstract of your paper to the main MSOM Conference, a decision on that submission will come separately.
Sincerely,
Healthcare Operations;SIG Co-Chairs
---------------------
Referee: 1
Please summarize the paper briefly. - Rev: Please see the attached report.
Referee: 2
Please summarize the paper briefly. - Rev: The paper examines how governments should combine biosecurity measures and reporting incentives to control African swine fever. Using a game-theoretic model, it shows that compensation encourages reporting, while stronger biosecurity may reduce it. The optimal policy sets target levels for both and adjusts compensation accordingly, with flexibility needed depending on reporting costs and farm heterogeneity.-
Referee: 1
Comments to the Author (old)
(There are no comments.)
Referee: 2
Comments to the Author (old)
(There are no comments.)-
Reply all
Reply to author
Forward
0 new messages