How the numbers are made
Every number on this site is one of four classes: a live FRED pull refreshed nightly, an external reference on its own cadence (some pulled, some pending), an editorial constant baked into the model, or an inline TBD waiting for a citation. §1 covers the FRED layer; §2 covers everything else we cite.
FRED snapshot generated · Jun 8, 2026, 6:03:07 PM
Refresh cadence · backend ingests FRED nightly at 02:00 UTC; the frontend rebakes the static snapshot at 02:30 UTC. Non-FRED sources refresh on their own cadence — see §2.
§1 · Live FRED data
Each row is a FRED series the essay uses, refreshed nightly by the backend ingest. Two dates per row: as of is the period the data point describes (e.g. CPI for May 2026); pulled is when our backend last fetched it. They are usually a few days to a few weeks apart depending on release cadence.
| Series | Value | As of | Pulled | Used for |
|---|---|---|---|---|
Risk-free rate (3-month T-bill) 3-month Treasury bill rate (TB3MS via FRED), converted from discount basis to bond-equivalent yield using BEY = (discount × 365) / (360 - discount × days) | 3.68 | May 1, 2026 | 2026-06-08 02:00 UTC | Short-end of term structure; §3 risk-free baseline; calculator fair-comp tier |
Real risk-free rate (10-year TIPS) 10-year Treasury Inflation-Protected Security yield via FRED | 2.04 | May 1, 2026 | 2026-06-08 02:00 UTC | Real-rate baseline for the TIPS-equivalent tier; inflation-projection anchor |
1-year Treasury yield 1-year Treasury constant-maturity yield via FRED | 3.79 | May 1, 2026 | 2026-06-08 02:00 UTC | Term-structure DCF: discount + drag rate for ≤1y horizon |
2-year Treasury yield 2-year Treasury constant-maturity yield via FRED | 4.00 | May 1, 2026 | 2026-06-08 02:00 UTC | Term-structure DCF: 2y interpolation node |
10-year Treasury yield 10-year Treasury constant-maturity yield via FRED | 4.48 | May 1, 2026 | 2026-06-08 02:00 UTC | 10y interpolation node; headline aggregate-rate reference in §6 prose |
30-year Treasury yield 30-year Treasury constant-maturity yield via FRED | 5.03 | May 1, 2026 | 2026-06-08 02:01 UTC | Long-tenor extrapolation; multi-decade DCF horizon |
Needs basket inflation Composite of BLS CPI components via FRED, weighted by shadowstats | 250.99 | Apr 1, 2026 | 2026-06-08 02:00 UTC | §3 inflation pass-through; mean-reverting projection initial condition |
Total US wages and salaries BEA NIPA, Compensation of employees - wages and salaries, via FRED | 12,981.39 | Jan 1, 2025 | 2026-06-08 02:01 UTC | §6 aggregate denominator (US private-sector wage bill annualized) |
Payroll lag tax Derived: BEA wages × (14-day average lag / 365.25) × 3mo T-bill mean. Fair compensation view. | 20.73 | Jan 1, 2025 | 2026-06-08 02:01 UTC | §1 headline + §6 aggregate annual foregone interest (computed downstream) |
All series are pulled via the FRED API. The needs-basket row is an editorial reweighting of CPI components, not a single FRED series — see Editorial constants below. payroll-lag-tax is a derived series our backend computes from total wages plus the term-structure Treasury curve, not pulled directly.
§2 · Pending / external data
Non-FRED sources the essay cites. None auto-refresh today; each has its own cadence and pull status. We surface them here so the reader can tell which TBDs in the prose mean "the data exists, we haven't wired it yet" vs. "the data does not exist in a form we can pull."
| Series | Source | Cadence | Status | Used for |
|---|---|---|---|---|
| Investment-grade corporate spread (OAS) | FRED · BAMLC0A0CM (ICE BofA US Corp Master) | Daily | Static · cited | §3 mirror-image pricing: IG firm commercial-paper / revolver replacement cost — ~76 bps as of May 2026 |
| High-yield corporate spread (OAS) | FRED · BAMLH0A0HYM2 (ICE BofA US High Yield) | Daily | Static · cited | §3 mirror-image pricing: HY firm working-capital replacement cost — typically 300-500 bps non-stress |
| Payday loan fees + APR + borrower count | Center for Responsible Lending · Down the Drain (Feb 2025); CFPB Data Points | Annual / irregular | Static · cited | §5 payday-loan fees (~$2.4B), APR (~391%), borrowers (~12M) |
| Overdraft / NSF Fee revenue | CFPB Data Spotlight: Overdraft/NSF Revenue (Apr 2024); Financial Health Network 2024 aggregate | Annual | Static · cited | §5 overdraft fees (~$12.1B in 2024; was $5.83B in 2023 before bank fee reversals) |
| BNPL volume + usage + late fees | CFPB BNPL Market Report (Dec 2025) | Periodic | Static · cited | §5 BNPL framing: 53.6M users, $45.2B originations in 2023; late fees ~0.18% of GMV (small) |
| FedNow transaction volume + FI adoption | Federal Reserve · quarterly publication | Quarterly | TBD · pull | §8 daily-settlement uptake claim |
| 11 USC §507(a)(4) priority unsecured wages cap | US Bankruptcy Code · Judicial Conference triennial CPI adjustment | Triennial (next 2028) | Static · cited | §3 "unsecured" unpacking: $17,150 since Apr 1, 2025 (was $15,150) |
| Damodaran implied ERP | NYU Stern · damodaran.com | Annual (January) | Static · cited | Calculator equity-expectation tier; baked into the constant in §3 of this page |
| NY Labor Law §191 (weekly pay for manual workers) | NY State Legislature · statutory text | Static | Cited | §8 cadence-floor precedent |
| Worker pay-frequency preferences | ADP Research · More Paydays, More Fairness (HR Experience Survey, 20,000 working adults, June 2022 – June 2023) | Periodic | Static · cited | §5 worker-side signal: 48% paid biweekly (modal); 43% unhappy with their pay frequency; preference for more frequent pay skews lower-income, female, younger |
| Organization-level intent to shift pay cadence | ADP Research · Potential of Payroll 2024 (1,825 payroll leaders in 20 countries, Aug-Sep 2024) | Periodic | Static · cited | §8 employer-side signal: 43% of organizations are exploring more frequent pay cycles |
| Lower-income worker / high-cost-credit usage correlation | CFPB Making Ends Meet in 2024 (Nov 2024); CFPB Data Spotlight: Developments in the Paycheck Advance Market (2024) | Annual | Static · cited | §5 closing paragraph — earned-wage-access users skew under federal poverty line, >80% hourly/gig, paying effective ~109.5% APR |
Status legend · TBD · wire means the series is available on FRED but the backend ingest doesn't pull it yet (low-effort fix); TBD · pull means the source is identified but requires a manual fetch (no public API, or just not wired); Static · cited means a single value is embedded and refreshed manually if at all; Cited means a text reference with no number flowing into the model.
§3 · Editorial constants
These values are not measured — they are chosen. Each is an editorial position that a reader could reasonably disagree with. We surface them so the disagreement can be specific.
| Constant | Value | Rationale |
|---|---|---|
| Employer-held float fraction | 75% / 25% | Downstream-allocation constant. Of the gross float that workers forgo, roughly 75% becomes employer working capital; the remaining ~25% is held briefly by the IRS / state via EFTPS withholding before remittance. Headline worker-loss numbers report the gross float and are NOT scaled by this fraction; the split is a separate analysis of where the foregone interest ultimately accrues. See "Where the float lands" below. |
| Editorial credit spread | 30 bps | Credit-spread premium added to the risk-free rate in the fair-compensation tier of the calculator. Reflects the worker bearing unsecured-creditor risk on accrued wages. |
| Fed inflation target | 2.0% | Long-run mean-reversion anchor for the inflation projection. Tracks the FOMC stated target rather than the current observation. |
| Inflation half-life | 5 years | Speed at which observed inflation reverts to the Fed target. Mid-range of standard term-structure inflation models. |
| Damodaran ERP | 5.0% | Equity risk premium for the upper-bound equity-expectation tier of the calculator. Tracks Damodaran’s rolled implied ERP estimate; heavily caveated as an upper bound, not a fair-value comparison. |
| Calculator career horizon | 40 years | Default DCF horizon. Reader-adjustable on the essay page. |
| Calculator wage growth | 3.5% | Default nominal wage growth in the DCF projection. Reader-adjustable. A derived historical-CAGR series is the natural follow-up. |
| Calculator discount rate | 4.0% | Default fallback discount rate when the term-structure curve is unavailable. Reader-adjustable. |
| Bonus cadence — payout-delay defaults | 7 / 30 / 45 / 60 d | Suggested payout-delay defaults per cadence (paycheck / quarterly / semi-annual / annual). Bank-rail processing for paychecks is days; bonus accrual close + verification is weeks; annual bonuses paid March 1 for a December 31 close are ~60 days. All cadences use the same per-period DCF math — only the period length and lag fraction change. Reader-editable per their actual contract. |
| Forfeiture leave-probability bands | 5% / 10% per year | Editorial bands for the bonus-mode forfeiture callout. Reflects typical voluntary + involuntary turnover ranges for professional / salaried workers (BLS JOLTS reports separations broadly in this band, though firm- and industry-specific rates vary widely). Expected forfeiture = leave_prob × notional. Used only in the conditional-comp callout; not applied to headline interest-tax numbers. |
§4 · Where the float lands
Headline numbers on the essay page report the gross worker loss — what the workforce forgoes in interest on the full float, regardless of recipient. This section is the downstream-allocation question: of that gross loss, who actually holds the dollars during the lag?
Payroll withholding (federal income tax, FICA, state income tax) is deducted from gross wages at each pay run and remitted to the IRS / state on a short cycle — typically within days for large employers under EFTPS. That portion of the float sits briefly with the government, not the employer, so it is not employer working capital. The rest — wages plus any non-tax deductions held with normal cash-management latency — is genuinely employer-held.
| Component | Share | $/yr | Recipient |
|---|---|---|---|
| Gross worker loss | 100% | $20.7B | Workforce, in aggregate. The headline. |
| — Employer working capital | 75% | $15.5B | Employer balance sheet. The §3 mirror-image argument quantifies this portion as the firm's spontaneous interest-free working capital. |
| — Withholding float | 25% | $5.2B | IRS / state, briefly. EFTPS short-cycle remittance means this float is in transit rather than parked. The interest is still forgone by the worker, just not captured by the employer. |
Split uses the editorial 75% / 25% constant in §3. A more granular cohort estimate (small-employer vs. mid-cap vs. enterprise; varying state withholding cadences) is on the roadmap and listed under §5 "what we don't have."
§5 · What we don't have
Empirical gaps that would make the argument stronger but the data does not currently exist in a form we can pull at build time. Naming them is part of the credibility commitment.
- Pay-cadence distribution by income decile. BLS tracks pay-frequency-by-industry; cross-tabbing against income would let us quantify the regressive compounding in §5/§6 directly instead of citing the directional finding.
- Per-firm employer credit spreads. §3's mirror-image pricing claim uses range estimates from published commercial-paper and revolver pricing surveys. A cohort-level dataset would let the calculator price the firm-side savings, not just the worker-side loss.
- Causal identification on cadence → high-cost borrowing. The literature linking pay-frequency to payday-loan demand is correlational with a few quasi-experimental wedges. §5 softens its verbs accordingly — see the essay.
- State-by-state §191-style legislation map. New York's weekly-pay mandate is well-documented; a current roll-up across states is not.
- On-demand-pay fee structures. DailyPay, Branch, EarnIn fees vary by employer contract and are not publicly disclosed at the contract level.
- Demographic modeling of the workforce. The population view projects today's $13T US wage bill forward at the assumed wage-growth rate. It does not model demographics: cohorts aging through careers, generational entry/exit, unemployment cycles, productivity-driven wage compression. It is a snapshot projection under today's pay-cadence conventions, not a cohort model. The personal calculator IS a cohort-of-one (one worker projected over a career); the population view assumes today's workforce composition persists and the wage bill grows at the assumed rate. Different math, different question. The headline annual number is robust; the multi-decade present value is for scale, not prediction.
- ADP-sample selection effects. ADP Research Institute data (cited in §5 and §8) covers roughly 1 in 6 US workers, which is the broadest available sample but skews toward firms that use ADP's services (typically mid-market and enterprise; smaller-employer representation is thinner). Directional findings on preference and intent are robust; precise quantitative splits should be read as "for the ADP-represented portion of the workforce" rather than a strict national average.
- Bonus structures and forfeiture data. The bonus cadences on the calculator (quarterly / semi-annual / annual) use the same per-period DCF math as paycheck cadences. We do not have a national-scale dataset on bonus structures — typical payout delays by industry, share of bonuses with forfeiture clauses, distribution of bonus size relative to base. Demographic skew matters too: bonus-as-meaningful-comp leans toward highly compensated employees (finance, law, consulting, tech), while hourly workers more often see smaller / more frequent commission structures. The calculator answers the individual question correctly; the population-scale story on delayed comp is on the roadmap and would benefit from BLS NCS-Employer-Costs or Compustat executive-comp data once we wire it.
- State-by-state bonus forfeiture law. Enforceability of bonus forfeiture clauses varies dramatically by state and by bonus structure (discretionary vs earned), and by separation cause (voluntary, for cause, or without cause). The essay cites California (AB 692, effective Jan 1 2026; CA labor law already treats earned bonuses as wages) and New York (S6775, 2025–26 session) as the load-bearing jurisdictions, but does not have a 50-state roll-up. The honest framing: the worst-case slice (discretionary bonus + voluntary departure + weak-protection state) is real but not universal. Many involuntary-without- cause cases in CA, NY, IL and other strong-protection states would survive the legal challenge — workers may have a claim they do not know about. A research-grade 50-state map of bonus-protection law against industry / structure is the natural next addition.