Updated 8 October 2026

Additional observations, fitted as points

Eight L3 applications now fit the published point estimates while preserving their exact L2 restrictions. The income and fiscal mappings remain fixed within each application.

One path-construction rule

Θ̂i,L = arg minθᵢ ∈ Θᵢ,ₗ [m̂i,L − gi,L(θᵢ)]′ Wi,L [m̂i,L − gi,L(θᵢ)],   L = 2, 3

The restriction set contains the anchor, logistic shape, calibrated support, attainment limits and credential ordering where relevant. L3 preserves the exact L2 observations and adds separately reported point targets. Within each of these eight cases, W is the identity matrix; standard errors do not set fitting weights or acceptance bands.

Exact matches minimize the criterion when attainable. These eight searches each retain one best-fitting path with nonzero residuals. This is a conditional fit, not proof of unique identification or global optimality. Five other L3 applications retain exact-match path ranges.

The eight updated L3 results

PolicyEarnings NPVWTPNet public costMVPFFit RMSE
E01 Florida Student Access Grant$13,837.52$11,267.50$1,027.2510.970.01024 years
E05 Wisconsin Scholars Grant$3,583.99$8,975.47$6,376.501.410.01589 years
E06 Kalamazoo Promise$54,197.83$58,802.48$16,289.063.610.01245 years
E17 Public-college spending$8,581.03$7,573.87$300.1725.230.00111 years
E19 Texas community-college tuition$13,120.84$10,879.96−$83.94Self-financing0.00388 years
E21 Tuition caps and freezes−$2,985.16−$1,205.97$1,146.07-1.050.00214 years
E37 Tennessee Promise$113,936.10$86,437.02−$14,831.39Self-financing0.35251 pp
E41 Cannabis legalization: education channel−$3,421.32−$2,354.48$47.02-50.080.00609 years

Amounts use each case’s registered monetary year and policy unit, so dollar levels are not cross-policy comparisons. Each row pairs WTP and net cost from the same fitted path. Self-financing means positive WTP with nonpositive net public cost. Tennessee retains its disclosed public-data income reconstruction. Cannabis values education only. RMSE includes the exact retained inputs; “pp” means percentage points.

Inspect every point target

Residual = fitted value − target. Source values retain their reported precision and the paper’s stated outcome and age conversions.

E01 · Florida Student Access Grant

Units: schooling-equivalent years · Monetary price year: 2000

InputTargetFittedResidualRole
source lag 10.0457500.056091+0.010341Additional point
source lag 20.1110830.105024-0.006060Additional point
source lag 30.1602920.146835-0.013457Additional point
source lag 40.1819170.181917-0.000000Exact L2
first semester age18.50.0147920.028925+0.014133Additional point
E05 · Wisconsin Scholars Grant

Units: schooling-equivalent years · Monetary price year: 2009

InputTargetFittedResidualRole
cohort1 cumulative credits0.0375000.068428+0.030928Additional point
Annual credit flow 10.0208330.025938+0.005104Additional point
Annual credit flow 20.0083330.023708+0.015375Additional point
Annual credit flow 30.0041670.018782+0.014615Additional point
Annual credit flow 40.0166670.037330+0.020664Additional point
Annual credit flow 50.0416670.027938-0.013729Additional point
Annual credit flow 60.0208330.016076-0.004757Additional point
pooled cohorts23 cumulative credits0.0875000.073306-0.014194Additional point
Retained L2 aggregate0.0717720.071772-0.000000Exact L2
E06 · Kalamazoo Promise

Units: schooling-equivalent years · Monetary price year: 2012

InputTargetFittedResidualRole
Retained L2 aggregate0.2733330.273333+0.000000Exact L2
two year credits0.1348330.122854-0.011980Additional point
three year credits0.1795420.197463+0.017922Additional point
E17 · Public-college spending

Units: schooling-equivalent years · Monetary price year: 2001

InputTargetFittedResidualRole
Retained L2 aggregate0.0460130.046013+0.000000Exact L2
institution event t00.0050410.006235+0.001194Additional point
institution event t10.0131990.011239-0.001960Additional point
institution event t20.0140110.013845-0.000166Additional point
institution event t30.0137620.014693+0.000931Additional point
E19 · Texas community-college tuition

Units: schooling-equivalent years · Monetary price year: 2012

InputTargetFittedResidualRole
source age 180.0315500.038841+0.007291Additional point
source age 190.0775500.078264+0.000714Additional point
source age 200.1127500.114291+0.001541Additional point
source age 210.1456500.144871-0.000779Additional point
source age 220.1721500.169401-0.002749Additional point
source age 230.1872500.188188+0.000938Additional point
source age 240.2094500.202006-0.007444Additional point
source age 260.2184500.218450+0.000000Exact L2
E21 · Tuition caps and freezes

Units: schooling-equivalent years · Monetary price year: 2001

InputTargetFittedResidualRole
Retained L2 aggregate0.0004610.000461+0.000000Exact L2
institution event t0-0.0009210.001154+0.002075Additional point
institution event t10.0037990.000994-0.002805Additional point
institution event t20.001785-0.000126-0.001910Additional point
institution event t3-0.004202-0.001562+0.002640Additional point
E37 · Tennessee Promise

Units: percentage points · Monetary price year: 2021 (reconstruction convention)

InputTargetFittedResidualRole
ce:C 18 193.5000003.676633+0.176633Additional point
ce:C 19 203.2000002.680707-0.519293Additional point
ce:C 20 212.5000002.950924+0.450924Additional point
ce:C 21 222.7000003.334780+0.634780Additional point
ce:C 22 232.8000003.467326+0.667326Additional point
ce:C 23 243.5000003.500000+0.000000Exact L2
aa:A 18 190.2000000.755623+0.555623Additional point
aa:A 19 201.4000001.625603+0.225603Additional point
aa:A 20 213.1000002.517882-0.582118Additional point
aa:A 21 222.6000002.836830+0.236830Additional point
aa:A 22 232.9000002.927979+0.027979Additional point
aa:A 23 244.0000004.000000-0.000000Exact L2
ba:B 18 19-0.100000-0.027279+0.072721Additional point
ba:B 19 200.100000-0.076261-0.176261Additional point
ba:B 20 210.001000-0.140953-0.141953Additional point
ba:B 21 22-0.2000000.027986+0.227986Additional point
ba:B 22 231.2000001.057222-0.142778Additional point
ba:B 23 242.7000002.700000+0.000000Exact L2
E41 · Cannabis legalization: education channel

Units: schooling-equivalent years · Monetary price year: 2000

InputTargetFittedResidualRole
BA four year bridge-0.032000-0.026976+0.005024Additional point
BA five year bridge-0.022000-0.031273-0.009273Additional point
BA six year bridge-0.036000-0.036000+0.000000Exact L2

Wisconsin’s semester and cumulative effects do not form an established exact adding-up identity. The original values and cohort windows are retained; residuals can reflect aggregation and model approximation.

Input uncertainty remains separate

Public-college spending, Texas tuition, tuition caps, Tennessee Promise and cannabis each refit 1,000 registered input perturbations. Tuition caps has 1,000 verified monetary fits and 998 defined MVPFs. Florida, Wisconsin and Kalamazoo lack complete joint L3 draws; no full-L3 uncertainty interval is imputed. The paper reports the inputs varied and the limits of these conditional summaries.

The interactive explorer continues to illustrate Chapter 3’s exact, generated observations. These policy results use the source-specific observation operators and accounts documented in the paper.