# Money, prices, wages, houses and Bitcoin: research record

Snapshot: 3 November 2026. Article: `/journal/the-wage-earners-gap`.

## Question

How did the dollar supply, consumer prices, hourly pay and the price of a house move against each other since 1964, what does an hour of work buy now against then, and where does Bitcoin's price sit against the power-law trend?

## Files

- `request.json`: the dataset declaration (FRED series, the mempool.space price file, rights as verified).
- `raw/<ID>.csv`, `raw/<ID>.html`: FRED responses exactly as served, fetched by `scripts/research/fetch-fred.mjs` in GitHub Actions. `raw_snapshot.json`: every observation as its source string, each file's SHA-256, the header fields and the retrieval time.
- `raw/mempool-historical-price-usd.json`: the Bitcoin price history from mempool.space's public API, fetched by `scripts/research/fetch-urls.mjs` in the same run. The API serves one line of 1,120,469 bytes; the file is stored indented, and compacting it reproduces the served bytes. `raw_files.json`: URL, retrieval time, bytes and SHA-256 as served, and the stored form's bytes and SHA-256. The build verifies both.
- `wordmark.png`: the Loop XXI wordmark embedded in each chart as the watermark.
- `build.mjs`: offline rebuild. Validates the inputs, computes every derived number, writes `wage-gap.json`, `validation.json`, the six chart SVGs and `sha256.json`. `node build.mjs` in this directory; no network, no dependencies.
- `scripts/research/chart.mjs`: the chart renderer (log scale and the logo were added for this record). `scripts/research/rasterize.py`: SVG to PNG (2x) and WebP with the site font.
- Tests: `scripts/research/wage-gap.test.mjs`, `scripts/research/chart.test.mjs` (in `npm test`).

## Series

| ID | Title | Units | Frequency | Adjustment | Range in snapshot | Last updated (FRED) |
| --- | --- | --- | --- | --- | --- | --- |
| M2SL | M2 | Billions of dollars | Monthly | Seasonally adjusted | 1959-01-01 to 2026-09-01 | 2026-10-27 12:01 PM CDT |
| CPIAUCSL | Consumer Price Index for All Urban Consumers: All Items in U.S. City Average | Index 1982-1984=100 | Monthly | Seasonally adjusted | 1947-01-01 to 2026-09-01 (one missing month, left missing) | 2026-10-14 8:08 AM CDT |
| AHETPI | Average Hourly Earnings of Production and Nonsupervisory Employees, Total Private | Dollars per hour | Monthly | Seasonally adjusted | 1964-01-01 to 2026-09-01 | 2026-10-02 8:26 AM CDT |
| MSPUS | Median Sales Price of New Houses Sold for the United States | Dollars | Quarterly | Not seasonally adjusted | 1963-01-01 to 2026-07-01 | 2026-10-27 9:03 AM CDT |

Sources: Board of Governors of the Federal Reserve System (M2SL); U.S. Bureau of Labor Statistics (CPIAUCSL, AHETPI); U.S. Census Bureau and U.S. Department of Housing and Urban Development (MSPUS); all via FRED, Federal Reserve Bank of St. Louis. Retrieved 2026-11-03T19:16Z. Vintage: the values published as of that retrieval; later retrievals can differ.

Bitcoin price: mempool.space public API, `https://mempool.space/api/v1/historical-price?currency=USD`, retrieved 2026-11-03T19:16:54Z, 34,836 points from 2010-07-19 to 2026-11-03 19:00 UTC (hourly for the recent period, weekly earlier). The chart draws at most one point per three days plus the last point.

## Rights

- FRED tags each of the four series "Public Domain: Citation Requested" (verified on each series page on 2026-11-03; URLs in `request.json`). Every chart carries the requested citation. FRED's page banner "Data in this graph are copyrighted" refers to FRED's own graph images, which we do not reuse.
- mempool.space price history: the provider declares no rights for this endpoint as of 2026-11-03. The series is a compilation of exchange prices; we treat it as facts, attribute it on the chart and in this record, and do not mirror it for download beyond this folder.
- Gold and stock price series were not charted: the public-domain sources above do not carry them, and the sources we checked for an earlier record (Coin Metrics Community, Yahoo, FRED's S&P and ICE gold series) are licensed or restricted.

## Formulas

- Index: `100 * value / value at 1964-01-01`, monthly; the quarterly house price at each quarter's first month.
- Hours of work per median new house: `MSPUS (quarter) / AHETPI (first month of the quarter)`.
- Hourly earnings in January 1964 dollars: `AHETPI * CPIAUCSL(1964-01) / CPIAUCSL(month)`.
- Power-law trend: `price = 1e-17 * days^5.8`, days since 2009-01-03T00:00Z, as proposed by Giovanni Santostasi and reported by Decrypt (URL in the article's sources). Past the last price point the dashed line is the same formula carried forward; it is a fit extended, not a forecast.
- Price to trend: `latest price / trend value on the same day`.

## Headline numbers (from `validation.json`)

- Since January 1964 to the latest observation: M2 ×59.25 (5,824.6%, to September 2026); median new house price ×21.26 (2,025.9%, to Q3 2026); average hourly earnings ×13.04 (1,204.0%, to September 2026); consumer prices ×10.87 (986.9%, to September 2026).
- Hours of work per median new house: 7,400 in Q1 1964; 12,131 in Q3 2026; peak 16,045 in Q2 2022.
- Average hourly earnings in January 1964 dollars: $2.50 in January 1964; $3.00 in September 2026; peak $3.04 in April 2020.
- Bitcoin: $77,506 on 2026-11-03; trend $131,897 the same day (price at 0.59 of trend); trend extended $190,018 at 2028-01-01, $339,747 at 2030-01-01, $737,350 at 2033-01-01, $1,459,589 at 2036-01-01.

## Limits

- The house series is the median new house sold, not all houses. The pay series covers production and nonsupervisory workers, not all employees. The CPI is one basket, used as published.
- The power law is a fit to past prices. A fit can fail; the extension is arithmetic, not a forecast, and nothing here is investment advice.
- The April 2020 peak in real hourly earnings reflects the mix of jobs that month, when low-paid work fell away, not a raise.

## Reproduce

```
cd assets/research/wage-gap-2026-11-03 && node build.mjs
python3 -I ../../../scripts/research/rasterize.py *.svg
```

`sha256.json` lists the hash of every input and output after a build. Code version: the git commit that last touched this directory.
