PASS escrow optimality: 6566 randomly drawn profile-contingent schemes (payments to committers and to investors who decline) implement formation by iterated strict dominance; none has liability at or below R*
PASS escrow optimality: the escrowed schedule at R*+eps implements formation robustly in every drawn structure (254 of 254)
PASS fragile example: no action dominated, no fully mixed equilibrium (discriminant -1.16), liability 2.25 against R* = 3
PASS fragile example: the largest probability of failure over correlated equilibria is 813/815 = 0.9975
PASS correlated robustness: in 40 closed cohorts, the least liability over certificates of unique correlated equilibrium is never below R* (34 within 0.15 of it)
PASS correlated robustness, general form: over certificates of unique correlated equilibrium the least liability is never below R* in 30 counts (closed cohorts and seat counts, with outsiders) nor below R_c in 30 random structures
PASS correlated example beyond counts: liability 6.5 < R* = 7 (R_c = 6); unique correlated equilibrium with gains 20 (exact dual certificate, weights 1,1,6,6, and primal LP); with gains 14 a correlated equilibrium fails for certain; no action dominated
PASS correlated example: with gains of 200 a scheme of liability 6.041 makes formation the unique correlated equilibrium (primal LP)
PASS correlated example with a payment to a decliner: 3 of 4 seats, liability 1.1 < R* = 2, unique correlated equilibrium (primal LP)
PASS portfolio of promises: one capacity slightly above max_k R*_k, honored by seniority, implements formation of every investment robustly (120 of 120); separate promisors need the sum
PASS staged escrow implements formation by backward iterated dominance with liability R-hat+eps in 300 of 300 random general structures
PASS staged-escrow lower bound: 16952 random schemes implement formation by backward iterated dominance; none has liability below R-hat (or equal when R-hat>0)
PASS a structure with R-hat = 0 is implemented by backward iterated dominance with no payment
PASS one hundred equal investors and ten rounds: staged 9h, time-invariant 90h, simultaneous 99h
PASS normal form: refusing is a best response to the plan of refusing after a commitment, and that plan is a best response to refusal (failure rationalizable)
PASS screening: escrow with a premium (200/200) or a fee (200/200) yields willing formation as the unique ICR outcome with unwilling losses near zero; without a charge, none (0)
PASS screening: a fee above the willing gain (7 > 1) fails although it lies in ((1-p)(beta-h), beta-h); a fee of 2/5 below the gain and the margin, with beta = 3/2, works
PASS order with escrowed payments equal to expenses: willing formation at every history whatever the gains and losses (300/300)
PASS calibration: h = $100k (3.3 bp); R* = $2.9m (97 bp) with all 30, $0.9m with 10 or with a staged escrow over three public rounds; holding cost $14.5k at 50 bp, $290k (9.7 bp) at 10%
PASS calibration: break-even expense share for organizational expenses 40% (50 bp capacity cost) or 2% (10%); for management fees at or above the commitment itself
PASS calendar or capacity: the staged escrow holds (n - l)h in total over l rounds for every partition (200 random); threshold carry of one investor for a month is $42 at 50 bp and $833 at 10%
PASS private willingness: simultaneous screening escrow pays 725,869 (24.2 bp) and wastes 750,899; an order wastes 357,865 (11.9 bp) and needs no payment once gains exceed 33,838; formation probability 0.740
PASS blind pool: deal-by-deal surplus n p^n g - W_n confirmed by exact enumeration (n = 1..6); D(2) = -0.002, D(3) = 0.263 in the example; D(1) = -(1-p)lambda and D positive and increasing from n-hat on in 300 random economies
23 checks; 0 failures
