US Infrastructure Model — Sources & Assumptions
This file keeps the model's data and assumptions apart from its framing,
following the repository's evidentiary discipline (see
../politico/SAKES.md and the rule that a model output
is a computation over assumptions, not a measurement). Nothing here is a
forecast or an official figure.
Status of the committed numbers
Every value in sectors.csv and the defaults in
model.sleela is an ILLUSTRATIVE_PLACEHOLDER. They are round
numbers chosen to exercise the model and to be plausible in order of magnitude —
not sourced measurements. The provenance column in sectors.csv records this
explicitly; replace it with a citation (author, publication, date, table) when a
figure is backed by a primary source, and only then treat that row as data.
The three layers (what can and cannot be trusted)
| Layer | Items | Trust |
|---|---|---|
| Identity (true by definition) | Y = C + I + G; renewalNeed = assetBase × depreciation; backlogDelta = renewalNeed − annualInvest; gap = max(0, backlogDelta) | Exact given the inputs. |
| Behavioral assumption (hypothesis) | output multiplier; per-sector bcRatio; discountRate | Defensible only within a cited range; results are sensitive to these. |
| Data (must be sourced) | assetBase, annualInvest, depreciation, conditionIdx | Placeholder until a dated primary source is attached. |
Where real figures come from (to be filled in per row)
When sourcing the placeholders, draw each quantity from a named primary release and record the retrieval date. Suitable primary sources for US infrastructure economics include (consult and cite the specific table/vintage — do not treat this list as data):
- BEA — National Income and Product Accounts (GDP components
C,I,G) and Fixed Assets Accounts (asset stocks, depreciation). - Census Bureau — Annual Value of Construction Put in Place (public/private investment by category).
- CBO — public spending on transportation and water infrastructure.
- DOT / FHWA, EIA, FCC — sector-specific asset condition and investment series for transport, energy, and broadband respectively.
- ASCE Infrastructure Report Card — condition proxies and investment-gap
framing (note: an advocacy source; treat its gap figures as claims to test,
per
METAPHOR ≠ FACT).
Assumption ranges (behavioral layer)
Documented so a reader can vary them deliberately rather than accept the defaults:
- Spending multiplier — public-infrastructure fiscal multipliers are commonly
discussed in a wide band (roughly 0.5–2.0+ depending on slack, financing, and
horizon). The model's default
1.5is a mid-range illustrative choice. - Benefit–cost ratio (
bcRatio) — appraisal BCRs for maintenance and targeted expansion frequently exceed 1; the per-sector values here (2.2–3.8) are illustrative and vary widely by project. - Discount rate — the default
0.03(3% real) is an illustrative long-horizon public-appraisal rate; sensitivity to this choice should be reported.
Reproducing / extending
- Replace
sectors.csvvalues with sourced figures and update theprovenancecolumn per row. - Mirror the sourced inputs into the sector definitions in
model.sleela(the program is self-contained; the CSV is the human-readable companion). - Re-run and report results with the assumption values stated inline, so the output is never mistaken for a measurement.
Per project discipline: the existence of a computed number is not proof of a claim. Cite the source, state the assumptions, and keep interpretation separate from the identity-level arithmetic.