Research brief: risk aversion, loss aversion, and post-failure decision patterns in GC and trades-business decision-makers (May 2026)

Status: Companion brief to [[research-brief-psychology-gc-marketing-aversion-may-2026]]. Excavates the cognitive mechanics underlying the risk model Ontario residential/ICI GC owners bring to vendor decisions. Compiled May 2026 (~9,500 words). Internal reference; not customer-facing.

TL;DR

• Loss aversion is real but smaller than folklore says. Brown-Imai-Vieider-Camerer 2024 meta-analysis of 607 estimates finds mean λ = 1.955, CI [1.820, 2.102] — below the canonical Tversky-Kahneman 1992 λ = 2.25 that has dominated marketing discourse. See [[brown-imai-vieider-camerer-2024-meta-analysis-lambda-1955]].
• The reference point is the draw cycle, not annual revenue. Köszegi-Rabin 2006 reference-dependent preferences + Ontario's 28/7-day prompt-payment cycle = marketing/software invoices are evaluated against the current draw's net position. See [[koszegi-rabin-2006-reference-dependent-preferences]] and [[ontario-construction-act-2026-7-day-deeming-rule]].
• The fourfold pattern is what Candid sees in sales. Risk-averse for moderate-probability gains (the marketing upside), risk-seeking for moderate-probability losses (doubling down on a stalled in-house build). See [[fourfold-pattern-risk-preferences-gc-cognitive-signature]].
• Expert-entrepreneur cognition runs on affordable loss, not expected return. Sarasvathy's effectuation research is the load-bearing decision-rule research for this buyer. See [[sarasvathy-effectuation-bird-in-hand-affordable-loss]].
• Owner-operators are loss-averse on dollars and overconfident on execution. The Camerer-Lovallo / Cooper-Woo-Dunkelberg / Busenitz-Barney pattern explains why a GC rejects a vendor pitch on financial-risk grounds and simultaneously believes his nephew can build the same website in a weekend. See [[overconfidence-in-house-comparator-pattern-for-gc-pitches]].
• "Once burned" is asymmetric Bayesian updating + negativity dominance, not recency bias. Kim-Ferrin-Cooper-Dirks 2004 + Rozin-Royzman 2001. See [[kim-ferrin-cooper-dirks-2004-competence-vs-integrity-trust-repair]] and [[rozin-royzman-2001-negativity-dominance]].
• Scarcity collapses planning horizons. Mullainathan-Shafir 2013 bandwidth tax. For a GC mid-cycle on payroll, a 12-month payback is cognitively distant. See [[mullainathan-shafir-2013-scarcity-bandwidth-tax]].
• Bounded-loss frames work; implicit-loss frames backfire. Follows from prospect theory + affordable-loss decision rule. Refinement of the May 2026 foundation brief's loss-framing recommendation. See [[bounded-vs-implicit-loss-frame-distinction]].
• Social proof works only when the reference group is proximate. Goldstein-Cialdini-Griskevicius 2008 "provincial norms" outperform generic descriptive norms. A SaaS case study is not social proof for a Cambridge framer. See [[goldstein-cialdini-griskevicius-2008-provincial-norms-hotel-towel]].
• Trust repair after competence violation is achievable; after integrity violation it is not. Apologize for competence breaches; deny culpability for integrity breaches. See [[kim-ferrin-cooper-dirks-2004-competence-vs-integrity-trust-repair]].
• The Construction Act's January 1 2026 amendments are reshaping cash-flow cognition. Mandatory annual holdback release + 7-day deeming + continued 28/7 prompt-payment accelerate but do not eliminate the structural mismatch driving short time horizons. See [[ontario-construction-act-2026-7-day-deeming-rule]] and [[ontario-construction-act-2026-mandatory-annual-holdback-release]].
• ODACC adjudication has grown 9x in five years but cannot yet be treated as a reliable backstop. Notices grew 32 (FY20) → 277 (FY24); 30-day statutory compliance fell to 65% in FY23. See [[odacc-five-year-operating-data-fy2020-fy2024]] and [[odacc-30-day-compliance-erosion-65pct-fy2023]].

The nine strategic implications (R1–R9)

1. [[rule-price-in-affordable-loss-terms-not-expected-return]]
2. [[rule-engineer-explicit-kill-criteria-into-engagements]]
3. [[rule-lead-with-certain-deliverable-not-probable-outcome]]
4. [[rule-surface-and-reframe-in-house-comparator]]
5. [[rule-diagnose-prior-vendor-failure-as-competence-or-integrity]]
6. [[rule-radius-matched-peer-social-proof-only]]
7. [[rule-bird-in-hand-pitch-architecture-default]]
8. [[rule-payment-terms-aligned-with-gc-draw-cycle]]
9. [[rule-replace-implicit-with-bounded-loss-framing]]

Sections

1. Behavioral-economics foundations — [[brown-imai-vieider-camerer-2024-meta-analysis-lambda-1955]], [[tversky-kahneman-1992-probability-weighting-certainty-effect]], [[tversky-kahneman-1981-framing-decisions-science]], [[koszegi-rabin-2006-reference-dependent-preferences]], [[fourfold-pattern-risk-preferences-gc-cognitive-signature]]; foundation entry [[kahneman-tversky-prospect-theory-loss-aversion-2to1]].
2. Entrepreneur-specific risk profiles — [[knight-1921-risk-uncertainty-distinction]], [[stewart-roth-2001-entrepreneur-risk-propensity-contested]], [[sarasvathy-effectuation-bird-in-hand-affordable-loss]], [[read-song-smit-2009-effectuation-meta-analysis-9897-ventures]], [[camerer-lovallo-1999-overconfidence-reference-group-neglect]], [[cooper-woo-dunkelberg-1988-entrepreneur-overconfidence-81pct-33pct]], [[busenitz-barney-1997-entrepreneur-vs-manager-overconfidence]], [[overconfidence-in-house-comparator-pattern-for-gc-pitches]].
3. "Once burned" pattern — [[tversky-kahneman-1973-availability-recency-vs-once-burned]], [[shefrin-statman-1985-disposition-effect-narrow-framing]], [[mayer-davis-schoorman-1995-trust-ability-benevolence-integrity]], [[kim-ferrin-cooper-dirks-2004-competence-vs-integrity-trust-repair]], [[rozin-royzman-2001-negativity-dominance]], [[doney-cannon-1997-b2b-trust-buyer-seller-relationships]], [[palmatier-2006-relationship-marketing-meta-analysis]].
4. Project-based revenue & time horizon — [[ontario-construction-act-2026-7-day-deeming-rule]], [[ontario-construction-act-2026-mandatory-annual-holdback-release]], [[odacc-five-year-operating-data-fy2020-fy2024]], [[odacc-30-day-compliance-erosion-65pct-fy2023]], [[odacc-claim-amount-sectoral-split-fy2024]], [[frederick-loewenstein-odonoghue-2002-hyperbolic-discounting]], [[mullainathan-shafir-2013-scarcity-bandwidth-tax]], [[mani-et-al-2013-sugarcane-farmers-cognitive-test]], [[chba-hmi-q4-2025-ontario-10pt4-record-lows]].
5. Persuasion architectures — [[cialdini-six-principles-audited-for-gc-buyer]], [[heath-brothers-success-framework-gc-fit]], [[bounded-vs-implicit-loss-frame-distinction]], [[goldstein-cialdini-griskevicius-2008-provincial-norms-hotel-towel]], [[berger-heath-2007-identity-signaling-de-persuasion]], [[bird-in-hand-sales-architecture-for-candid]], [[pitch-categories-pre-emptively-defeated-by-gc-cognitive-profile]].
6. Research gaps & caveats — [[risk-aversion-brief-research-gaps-may-2026]].

Relation to the May 2026 foundation brief

This brief refines two claims from the foundation brief:

• The foundation brief used the canonical λ ≈ 2:1 figure. Brown et al. 2024 puts the empirical mean at 1.955 — defensible as rough magnitude but not as a precise multiplier. Candid drafts should stop citing "losses loom twice as large" as if it were a precise constant.
• The foundation brief recommended explicit loss-framing. This brief refines that to bounded loss-framing: implicit ("you'll fall behind") fails on this audience; bounded ("here is the maximum you can lose, with kill points at month 2 and 4") works because it engages affordable-loss decision logic and respects the certainty effect. See [[bounded-vs-implicit-loss-frame-distinction]].