Research Agenda¶
This page lists the paper currently in submission preparation and a ranked set of follow-up paper ideas that can be built using the same underlying data (the BEC-SP centralized procurement panel 2016–2019, with the longer registry back to 2005; Pregão event-log drop-out bids; the March 2018 PGE-SP reinterpretation and the 2018 Decreto 9.412 cap shift; CMED pharmaceutical price ceilings; reference prices; plus linkages to RAIS firm/worker records and IBGE municipal panels). Each entry includes three target-journal options with my probability estimates of revise-and-resubmit (R&R) at each.
Probabilities are honest reviewer-style assessments; "Confirmed" is not an option for any single-jurisdiction paper from this data lake without cross-data replication. All probabilities assume a clean manuscript at submission.
Reading the R&R probability columns
R&R = invited to revise and resubmit (not Accept). Acceptance probability is roughly half of R&R for the listed top journals. "Strong fit" = my reviewer-side reading of the journal's recent publication mix; "Marginal fit" = stretch.
1. Current paper (in submission preparation)¶
The Price of Exclusion: SME Set-Asides in Public Procurement¶
Author: Darcio Genicolo-Martins (Insper)
Version: v8 (May 2026, JPubE submission) manuscript PDF · online appendix
Distinctive contributions:
- A price-formation decomposition of the set-aside — the price effect of an SME set-aside is split into a lost-discipline component (non-SMEs removed from the price-forming pool, SME pool held fixed: \(S_2 - S_1\)) and a protected-pool offset (the post-policy SME pool replacing the pre-policy one: \(S_3 - S_2\)). The set-aside price effect becomes a design question — how much is mechanical exclusion of rival bidders, and how much is the protected pool's response — rather than a single reduced-form number.
- Implementation via a reverse-auction setting — in São Paulo's Pregão (descending-clock electronic reverse auction), losing bidders' drop-out prices reveal type-specific willingness to supply under the maintained independent-private-values clock interpretation (Vickrey 1961; Milgrom–Weber 1982; Haile–Tamer 2003). The recovered primitives, an auction-level heterogeneity correction in the spirit of Krasnokutskaya (2011), and observed equilibrium entry simulate counterfactual price-formation objects under three pools.
- A recast of policy design — the relevant policy frontier is not SME support versus no support, but exclusionary redistribution versus support that preserves the price-forming bidder pool. A 10% SME price preference enters as a static design benchmark: it keeps the non-SMEs that discipline the price-forming order statistic but delivers less redistribution than full exclusion.
In standardized non-pharmaceutical procurement, the protected pool responds (SME participation roughly doubles) but does not replace the excluded discipline: the full set-aside generates a static welfare loss of 28.9% of the open-regime price at λ=0.30, with the exclusion component accounting for ~72% of the absolute price decomposition. The implied SME welfare weight required for a planner to prefer full exclusion to the 10% preference is 2.42. The exclusion-dominant ranking survives Turnbull winner-censoring (74%), strict-invariance of the post-policy SME distribution (85%), and replacement of the Poisson bidder-count process with the empirical class-period-type count distribution (69%). Pharmaceutical procurement is reported as a boundary case, not a second headline.
Target journals:
| Tier | Journal | R&R prob. | Fit rationale |
|---|---|---|---|
| 1 | JPubE (J. of Public Economics) | 30–40% | The natural home: SME-preference welfare analysis with an MCPF + Saez–Stantcheva frame is JPubE-core, and the policy-frontier recast is the kind of design contribution JPubE rewards. The binding constraints are single-jurisdiction external validity and the maintained IPV-clock assumption, which JPubE referees with a structural-IO bent will press hardest. |
| 2 | AEJ: Policy | 35–45% | Strong fit for the exclusionary-redistribution-vs-preference frontier and the implied-welfare-weight statistic; AEJ:Policy values clean policy-design counterfactuals tied to a real natural experiment (the March 2018 PGE-SP reversal). |
| 3 | IJIO (Int'l J. of Industrial Organization) | 45–55% | More reachable empirical-IO target; the structural recovery from drop-out bids and the entry-cost asymmetry fit the procurement-auction literature there. The welfare framing is less central than at JPubE/AEJ:Policy, but the auction mechanics travel well. |
Current submission gating constraints:
- Single-jurisdiction limitation — every estimate lives in São Paulo BEC. A second procurement jurisdiction (federal ComprasNet, or another state platform) replicating the open-vs-set-aside price effect would push R&R toward the upper end at every tier.
- Maintained IPV-clock assumption — the structural decomposition reads Pregão drop-outs as willingness-to-supply observations (H:ipv-clock-admissible). It is load-bearing and not testable within the data; the cross-modality GPV recovery from Convite first-price bids (AN-019) gives partial discipline, not proof.
- Static-only welfare — the paper measures the static price/welfare cost and does not credit any dynamic SME capacity-building benefit. Idea 2.1 below is the direct fix path.
2. Seven follow-up paper ideas using the same data¶
The seven ideas below are ordered by overlap with the current paper, from highest to none. Two extend the set-aside / price-formation framework directly (2.1–2.2); the rest take the BEC + RAIS + CMED + IBGE data lake into different research programs. The last two (2.6–2.7) are fully disjoint from the current paper — no set-aside, no price-formation decomposition, no set-aside-versus-preference frontier — and pose entirely different research questions (bureaucratic capacity; fiscal-year budget cycles) on the same data substrate.
Idea 2.1 — Do Set-Asides Build SME Capacity? The Dynamic Case for Exclusion¶
Overlap with current paper: High — directly answers the main limitation of the current paper (static-only welfare), using the same March 2018 shock and the same treated group.
Central question: The current paper measures only the static cost of the set-aside. The standard defense of set-asides is dynamic: the protected pool grows — SMEs that win build capacity, hire, and eventually compete unassisted. Does the data show this? Do SMEs that win under the set-aside expand employment, payroll, and survival relative to comparable SMEs that do not?
Empirical strategy: Link BEC SME winners to RAIS firm/worker trajectories (the linkage already built for AN-028). Event-study and Callaway–Sant'Anna designs around first set-aside win and around the March 2018 expansion of protected demand into Group 65. Outcomes: employment, wage bill, formal-sector survival, and subsequent unaided (open-auction) win rates. The dynamic benefit is then compared against the static welfare cost the current paper already quantifies.
Data reuse: BEC panel + RAIS linkage (already loaded), March 2018 shock. ~8–10 weeks; mostly RAIS panel assembly and event-study pipeline.
Target journals:
| Tier | Journal | R&R prob. | Rationale |
|---|---|---|---|
| 1 | J. of Public Economics | 30–40% | Pairs naturally with the current paper as the dynamic counterpart; JPubE would value a clean static-cost-vs-dynamic-benefit ledger on one policy. Identification of the dynamic margin is the cap. |
| 2 | AEJ: Applied Economics | 35–45% | RAIS-based firm-growth event study is squarely AEJ:A; the procurement-as-industrial-policy framing travels. |
| 3 | J. of Labor Economics | 30–40% | If the worker-flow / hiring margin is the lead (set-aside wins → formal hiring), JOLE is a strong secondary fit. |
Why this idea exists: It converts the current paper's headline limitation into a paper. If the dynamic benefit is small, it strengthens the current paper's policy reading; if large, it is the counterweight the static cost cannot see.
Idea 2.2 — Optimal SME Preference Design: Solving for the Margin, Sector by Sector¶
Overlap with current paper: Medium — same structural machinery (recovered primitives + BNE simulation), but pivots from evaluating one policy to designing the optimal one.
Central question: The current paper benchmarks a single 10% preference. What is the welfare-maximizing preference margin, and how does it vary across product groups with different SME/non-SME cost gaps and entry-cost asymmetries? Is there a sector where full exclusion is actually optimal under a defensible welfare weight?
Empirical strategy: Generalize the preference benchmark from AN-012 into a margin-choice problem. For each product cell, use the recovered type-specific cost distributions and calibrated entry costs (AN-030) to trace welfare as a function of the preference margin under a λ-grid and a grid of SME welfare weights. Map the policy frontier sector by sector.
Data reuse: Recovered primitives + entry costs already produced for the current paper; new code is the optimization sweep, not new data. ~6–8 weeks.
Target journals:
| Tier | Journal | R&R prob. | Rationale |
|---|---|---|---|
| 1 | AEJ: Policy | 35–45% | Optimal-design papers with an explicit planner objective and a real institutional preference instrument are a strong AEJ:Policy fit. |
| 2 | J. of Public Economics | 30–40% | JPubE likes mechanism-design-meets-welfare, but would want the structural assumptions stress-tested harder than a single-paper sweep allows. |
| 3 | IJIO | 45–55% | The auction-design mechanics fit IJIO; weaker on the welfare/planner side. |
Differentiator from current paper: the current paper evaluates two regimes (set-aside, 10% preference); this paper solves for the optimal preference and characterizes when exclusion can be rationalized.
Idea 2.3 — Procurement and Local Economic Development: Does Buying Local Pay?¶
Overlap with current paper: Low — same data substrate, but a development / fiscal-federalism question and outcomes (municipal employment), not auction welfare.
Central question: The current paper finds open auctions pull in geographically distant non-SME suppliers on high-value items (AN-003). Flip the welfare question: does keeping procurement local (which SME-only tendering tends to do) support local employment and formal-sector activity, and at what price premium? Is there a local-multiplier benefit that offsets part of the static price cost?
Empirical strategy: Municipality × year panel. Link BEC supplier locations (PBU and firm CEP) to RAIS municipal employment and IBGE fiscal/demographic data. Use the March 2018 shock and the 2018 Decreto cap shift as variation in how "local" procurement becomes. Outcomes: local formal employment, supplier diversity (HHI of who wins), local-supplier share, and the implied price premium for local sourcing.
Data reuse: BEC supplier geocoding + RAIS municipal + IBGE (IBGE/RAIS already in the monorepo). ~10–12 weeks; new spatial / municipal-panel pipeline.
Target journals:
| Tier | Journal | R&R prob. | Rationale |
|---|---|---|---|
| 1 | J. of Economic Geography | 50–60% | Spatial structure of public demand and local-supplier development is a perfect fit; smaller but high-impact within geography. |
| 2 | Regional Science and Urban Economics | 45–55% | Strong methodological fit for the local-multiplier-of-procurement question. |
| 3 | World Development | 40–50% | Brazilian municipal context with a clear policy lever; less competitive, values practical-policy work. |
Differentiator: the unit is the municipality, not the auction; the contribution is a local-development ledger, and the auction welfare machinery of the current paper is absent.
Idea 2.4 — Reference Prices and the Anchoring of Bids in Reverse Auctions¶
Overlap with current paper: None — same data, a market-design question about information rather than set-asides.
Central question: Every BEC tender carries a reference price (preço de referência) that bidders observe. Does that anchor shape bidding — compressing dispersion toward the reference, inviting a winner's-curse-style overshoot, or capping competition because no one bids far below it? Would changing reference-price disclosure raise competition or lower prices?
Empirical strategy: Bid-level analysis using the normalized bid \(c_\varepsilon = b/p^{\mathrm{ref}}\) already constructed for the structural sample. Exploit cross-cell and over-time variation in how reference prices are set (historical vs market-survey vs CMED-anchored) and any disclosure-rule changes. Test for bunching at the reference, dispersion compression, and the relationship between reference-price staleness and realized competition.
Data reuse: BEC bid-level + reference prices + CMED ceilings (all in hand). ~8–10 weeks.
Target journals:
| Tier | Journal | R&R prob. | Rationale |
|---|---|---|---|
| 1 | J. of Public Economics | 30–40% | Information design in procurement is JPubE-relevant; needs a sharp source of variation in reference-price rules to clear the identification bar. |
| 2 | AEJ: Applied Economics | 35–45% | Bunching / anchoring evidence on real auction data is a good AEJ:A fit. |
| 3 | Int'l J. of Industrial Organization | 45–55% | The auction-information mechanics fit IJIO well; more reachable. |
Differentiator: no set-aside, no SME framing; the object is the reference price as an information instrument in auction design.
Idea 2.5 — Do Price Ceilings Bind? CMED Caps and Pharmaceutical Procurement¶
Overlap with current paper: None — promotes the current paper's boundary case (pharmaceuticals) to the headline, with a different (health-economics / regulation) question.
Central question: Pharmaceuticals in BEC are subject to CMED federal price ceilings. Do those ceilings bind in procurement, and what is the pass-through? Does the ceiling act as a focal point that raises realized prices (a cap-as-floor effect), or does competition pull prices well below it? The current paper treats pharma as model-sensitive and sets it aside; this paper makes the CMED ceiling the object of study.
Empirical strategy: Restrict to CADMAT pharmaceutical classes (6531, 6532, 6536, 6581) where CMED ceilings apply. Compare realized winning prices to the binding CMED ceiling; exploit CMED ceiling revisions over time as variation. Test for bunching at the ceiling, cap-as-floor focal-point behavior, and heterogeneity by molecule competition (single-source vs multi-source generics).
Data reuse: BEC pharma sub-sample + CMED ceiling series + the within-Group-65 CMED split already built for AN-027. ~8–12 weeks; main lift is assembling the CMED ceiling panel.
Target journals:
| Tier | Journal | R&R prob. | Rationale |
|---|---|---|---|
| 1 | J. of Health Economics | 35–45% | Pharmaceutical price regulation and procurement is a JHE-core topic; the cap-as-floor question is novel for the Brazilian setting. |
| 2 | J. of Public Economics | 25–35% | Possible if the welfare/pass-through framing is sharp; JPubE prefers OECD pharma-pricing contexts and tighter identification. |
| 3 | AEJ: Policy | 30–40% | Price-ceiling design with a clear regulatory lever fits AEJ:Policy; needs a clean source of ceiling variation. |
Differentiator: a regulation paper about price ceilings, not a set-aside paper; the auction-welfare decomposition is absent and the pharmaceutical sector is the headline rather than the caveat.
Idea 2.6 — Who Buys Well? Bureaucratic Capacity and Procurement Performance Across 1,344 Buyers¶
Overlap with current paper: None — the unit of analysis and the object of study are the buyer, not the auction rule or SME eligibility.
Central question: Holding the good fixed, procurement outcomes vary enormously across buyer units. How much of the variation in prices paid, competition attracted, and tender success is attributable to the buyer — its organizational capacity and the people who run it — rather than to what is bought? Do more capable buyers obtain systematically lower prices and more bidders for the same item, and which observable capacity proxies predict effectiveness?
Empirical strategy: Decompose item-level outcomes (price relative to the reference, number of bidders, completion) into product, time, and buyer (PBU) components on the 1,344-PBU panel. Identify buyer effectiveness through a connected-set / movers design where the data support it (PBUs linked by shared items over time; if procurement officials can be tracked across PBUs, an AKM-style mover decomposition in the spirit of Best–Hjort–Szakonyi 2023). Relate the estimated buyer effects to capacity proxies (purchase volume, experience, centralization, turnover). Anchors: Bandiera–Prat–Valletti (2009, AER), Bandiera–Best–Khan–Prat (2021, QJE), Best–Hjort–Szakonyi (2023, AER).
Data reuse: BEC full panel (no new data); optional RAIS linkage if official identifiers are recoverable. ~10–12 weeks; the main lift is the connected-set / movers decomposition and the capacity proxies.
Target journals:
| Tier | Journal | R&R prob. | Rationale |
|---|---|---|---|
| 1 | J. of Public Economics | 30–40% | State capacity and waste in government spending is JPubE-core (the Bandiera–Prat–Valletti lineage). The binding cap is cleanly separating buyer effectiveness from buyer-specific selection of what they buy. |
| 2 | AEJ: Policy | 30–40% | Procurement-performance heterogeneity with a real policy lever (which buyers to support or centralize) fits AEJ:Policy. |
| 3 | JLEO (J. of Law, Economics & Organization) | 40–50% | The organizational-capacity angle and within-bureaucracy variation are a strong, more reachable JLEO fit. |
Differentiator: the contribution is a buyer-effectiveness ledger; the auction-welfare machinery of the current paper is entirely absent. It asks who procures well and why, not which rule prices the contract.
Idea 2.7 — Use It or Lose It: Fiscal-Year Cycles and Year-End Spending in Procurement¶
Overlap with current paper: None — a public-finance / budget-institutions question about when money is spent, orthogonal to set-asides and to price formation.
Central question: Annual budgets that do not roll over create an incentive to exhaust funds before year-end. Does São Paulo state procurement show the Liebman–Mahoney (2017) year-end spending surge — and is the rush wasteful? Do December tenders pay more relative to the reference price, attract fewer bidders, or complete worse than otherwise identical tenders earlier in the year? What is the implied efficiency cost of expiring budgets?
Empirical strategy: Use the precise BEC timestamps to build a within-year spending profile by PBU. Test for an end-of-fiscal-year volume spike and, conditional on the item, whether year-end tenders pay a premium over the reference price, draw fewer bidders, and differ in completion. Exploit cross-PBU variation in budget rigidity and the fiscal calendar; benchmark the realized year-end premium against the Liebman–Mahoney federal estimate (last-week spending ≈ 4.9× the rest-of-year weekly average; lower year-end quality). Anchor: Liebman–Mahoney (2017, AER).
Data reuse: BEC timestamps + reference prices + completion status (all in hand); no new data. ~8–10 weeks.
Target journals:
| Tier | Journal | R&R prob. | Rationale |
|---|---|---|---|
| 1 | J. of Public Economics | 35–45% | Use-it-or-lose-it wasteful year-end spending is squarely JPubE public finance; Liebman–Mahoney is the template, and a non-US, state-level replication with explicit price/competition/completion margins is a clean contribution. |
| 2 | AEJ: Policy | 35–45% | Budget-institution design (rollover rules) with a measurable efficiency cost is a strong AEJ:Policy fit. |
| 3 | National Tax Journal | 50–60% | Budget-cycle and public-budgeting work is a reachable, high-fit NTJ target. |
Differentiator: the identifying variation is the fiscal calendar, not the auction rule or SME eligibility; the contribution is a budget-institution efficiency cost, and the paper's auction-welfare decomposition is absent.
Summary: portfolio view¶
| Idea | Overlap | Top journal | Top R&R |
|---|---|---|---|
| Current paper | — | JPubE | 30–40% |
| 2.1 Dynamic capacity-building | High | AEJ: Applied | 35–45% |
| 2.2 Optimal preference design | Medium | AEJ: Policy | 35–45% |
| 2.3 Local economic development | Low | J. Economic Geography | 50–60% |
| 2.4 Reference-price anchoring | None | IJIO | 45–55% |
| 2.5 CMED price ceilings | None | J. Health Economics | 35–45% |
| 2.6 Bureaucratic capacity | None | JLEO | 40–50% |
| 2.7 Year-end spending | None | National Tax Journal | 50–60% |
The portfolio mixes:
- Highest R&R targets: National Tax Journal (Idea 2.7, ~50–60%) and J. of Economic Geography (Idea 2.3, smaller but high-fit), then IJIO (Idea 2.4) and JLEO (Idea 2.6), then AEJ:Policy / AEJ:Applied (Ideas 2.1, 2.2).
- Most coherent pairing: Idea 2.1 (dynamic capacity-building) is the direct counterpart to the current paper's static-only welfare and is the cleanest sequel.
- Cleanest reach: J. of Economic Geography (Idea 2.3), National Tax Journal (2.7), and the IJIO targets (2.2, 2.4) at ~50% with good positioning.
- Fully disjoint papers (2.4, 2.5, 2.6, 2.7) avoid cannibalizing the current paper's contribution, at the cost of more data assembly. The two newest (2.6 bureaucratic capacity, 2.7 year-end spending) are the cleanest intellectual breaks: they share only the data substrate, not the set-aside / price-formation frame.
Sequencing recommendation: complete the current paper's R&R cycle first; then pursue Idea 2.1 (Dynamic capacity-building) as the natural sequel — it answers the most predictable referee question about the current paper ("but don't set-asides build SME capacity over time?") and reuses the RAIS linkage already in hand. Idea 2.3 (Local development) is the clean break that uses similar data without competing intellectually with the current paper.
Last updated: 2026-05-25