Skip to content

Findings — The Price of Exclusion

Intuition (plain-language)

One idea ties the project together: in public procurement the price is set by the fiercest bidders, and SME set-asides work by removing them. Opening São Paulo's medical-supply tenders to all comers cut prices about 10–11% and drew in more — and more distant — firms; closing them again reversed it. The structural decomposition shows this exclusion, not any failure of SMEs to enter, does most of the damage (≈72% of the price effect), costing roughly R$38–89 million a year on one product group. The implied fix is not abandoning SME support but delivering it through a price preference that keeps the competitive pool intact.

This page is the curated index of what we have learned from the project's own analyses. Each finding is a claim about the world — the kind of statement that would go in the paper — and it may rest on several AN-NNN analyses, cited in its Sources footer. The detail, design, and caveats behind each number live on those analysis pages; this page is the synthesis.

This page is the directory of conclusions. Use docs/analyses/ to look up the per-result design, raw numbers, and full result tables; use this page to scan what we believe and at what confidence.

How to read the confidence tags

A traffic-light convention runs across both scales: 🟢 = strongest confidence, 🟡 = middle, 🔴 = weakest. The meaning of each color depends on whether the claim is empirical or interpretive.

Empirical findings — the color reflects the source of confidence, not the size of the effect:

  • 🟢 Replicated — the finding appears in multiple independent samples or studies that agree in direction and rough magnitude.
  • 🟡 Single source — one solid study, or one of our own runs, with no independent replication yet.
  • 🔴 Provisional — one descriptive cut that is parser-dependent, sensitive to sample definition, or flagged with a known caveat. Read the qualifier before quoting.

Interpretations — parallel scheme:

  • 🟢 Strong — multiple converging lines of evidence; alternatives have been considered and rejected.
  • 🟡 Plausible — consistent with the evidence but other readings remain open.
  • 🔴 Speculative — suggested by the data but unverified; flagged for follow-up rather than relied on.

Every finding here is currently 🟡: each rests on the project's own runs of the BEC structural sample, none has independent replication in another jurisdiction, and the structural pieces inherit the maintained IPV-clock interpretation of Pregão drop-outs (load-bearing for the decomposition; see H:ipv-clock-admissible).

Findings overview

Reduced-form policy effects (empirical)

Structural decomposition (interpretive, conditional on IPV-clock)

Static welfare and policy design (interpretive, conditional on IPV-clock + λ=0.30)

Scope (interpretive)

Open items for this page

  • Nothing is 🟢 yet. Every finding is single-source (our own run on the São Paulo BEC sample); promotion to 🟢 would require either an independent replication in another procurement jurisdiction or a second identifying source within São Paulo (e.g., a complementary structural recovery from Convite first-price bids that converges with the Pregão drop-out reading; partly addressed by the cross-modality check in §6).
  • The IPV-clock diagnostic battery has landed. What was an open threat-assessment item — "do the screens land clean?" — has been resolved on the differential question (does coordination intensify post-cutoff, the threat that would mechanically produce the exclusion-dominant decomposition through bid suppression). It does not: Conley-Decarolis close-pair shares are stable in non-pharma (16.9% → 16.8%) and fall in pharma (27.6% → 24.4%); Bajari-Ye T1 ratios fall in both classes (NP 2.63 → 1.83; PH 1.29 → 1.11). See AN-015 for the full screen battery. Three winner-censoring regimes give net \(S_3 - S_1\) in [0.246, 0.275] non-pharma and [0.308, 0.357] pharma — all positive, all large (AN-013); Gaussian-copula relaxation with within-auction cost correlation up to ρ_c = 0.3 drifts the exclusion share by <5 pp and the total effect by <10% (AN-014); strict invariance reinforces the dominance ordering, raising the exclusion share to 85% non-pharma and 79% pharma (AN-017); cross-modality GPV from Convite first-price aligns with Pregão drop-outs in the load-bearing pharma non-SME pre cell (\(c_{0.50}\) 0.712 vs 0.704; AN-019). The screens do not prove the IPV-clock reading — that is why findings remain 🟡 — but they rule out the most obvious bid-conduct deviations. Residual baseline clustering remains an acknowledged limitation (realized close-pair shares are 1.6–1.9× the null mean).
  • The 10% price preference is a static design benchmark, not a forecast. It holds entry and recovered willingness-to-supply primitives fixed; it does not solve a counterfactual legal regime in equilibrium. Any reading as a policy recommendation is interpretive and load-bearing on that scope choice.