The author holds that wages ought to be pegged to inflation, ahead of it and not behind it. This page tests the half a script can test: what the record says about wages and prices, and how much a peg that trails prices would have missed.
People ought to have wages pegged to inflation, ahead of it, not behind it, which, the author says, has historically been the case. That is a normative claim and this page does not derive it. (A “peg” here is indexation of wages to prices. It is not the PEG of WP-128, a gap measure; the two share a word.)
When the author sold investments in banks, the instruction was to tell customers to plan and invest for retirement and pay for insurance “because you never know.” The author’s point is that statistics let us know more than that. Sections 5 and 6 take that seriously: they put a range on what the record shows, and say where it cannot.
Tested: whether wages were behind prices historically, how often, which moves first, and what a peg to trailing inflation misses. Not tested: whether pegging wages would be good policy, and what indexation does to inflation itself (the wage–price question; section 7). Series: average hourly earnings of production and nonsupervisory employees (FRED AHETPI, from January 1964) against the all-items CPI (CPIAUCSL), 751 months from January 1964, retrieved live.
| Month | Real wage (Jan 1964 = 100) |
|---|---|
| Jan 1964 | 100.0 |
| Jan 1973 | 116.8 |
| Jan 1979 | 110.8 |
| Jan 1990 | 97.3 |
| Jan 2000 | 100.4 |
| Jan 2010 | 107.4 |
| Dec 2019 | 114.1 |
| Jun 2022 | 115.5 |
| Aug 2026 | 120.5 |
Over 62.5 years the real wage is up +20.5%, 0.30% a year. That is not “behind” in the long run. But the record has falls of 3% or more from a peak:
| Fall (peak → trough) | Depth | Back at the peak |
|---|---|---|
| Jan 1973 → Apr 1995 | 23.9% | Apr 2020 (567 months after the peak) |
| Apr 2020 → Jun 2022 | 5.2% | not yet |
The deepest is the one the author’s claim describes: the real wage peaked in January 1973, fell 23.9% by April 1995, and did not regain that peak until April 2020, 567 months later. A nominal raise of 23.9% at the trough would have restored it. The April 2020 level was itself lifted by whose jobs were still there that month, a known feature of that month’s data that this page does not compute; the second fall, 5.2% from there to Jun 2022, has not yet been recovered. A broader all-employee series, available only from March 2006, ends +12.6% above its start (the main series over the same months: +17.5%), so the direction does not depend on which series is used.
| Decade | Months below | Share |
|---|---|---|
| 1960s | 0 of 60 | 0% |
| 1970s | 55 of 120 | 46% |
| 1980s | 94 of 120 | 78% |
| 1990s | 59 of 120 | 49% |
| 2000s | 41 of 120 | 34% |
| 2010s | 26 of 120 | 22% |
| 2020s | 23 of 79 | 29% |
Across all 739 months with a 12-month comparison, wage growth was below inflation in 298 of 739 (40.3%). “Historically behind” is true of the 1970s to the 1990s (78% of the 1980s) and false of the 1960s (none) and the 2010s (22%).
Correlating 12-month wage growth at month t with 12-month inflation at month t+k, the best lag k is:
| Window | Best lag k (months) | r at best | r at −12 | r at 0 | r at +12 |
|---|---|---|---|---|---|
| 1965-1984 | 0 | 0.73 | 0.57 | 0.73 | 0.62 |
| 1985-2019 | -5 | 0.39 | 0.31 | 0.35 | 0.17 |
| 2020 on | -4 | 0.62 | 0.18 | 0.51 | 0.18 |
A negative k means inflation came first and wages followed. From 1985 to 2019 the best fit has wages following prices by about five months; before 1985 they moved together. Two persistent series at nearby lags are nearly flat in correlation, so this is a weak statement and is reported, not leaned on.
A peg that raises wages by the last twelve months’ inflation is behind by exactly the acceleration: inflation over the next twelve months minus inflation over the last twelve. Over 727 months that miss averages out (+0.10 points at the median, no bias), and it is large in the tails: the 90th percentile is +2.50 points and the worst month, Feb 2021, is +6.27 points short over the following year. So a trailing peg is not wrong on average. It is late, and late is expensive exactly when prices are moving. A peg ahead of prices needs a forecast; this page scores only the naive one and does not build another.
The acceleration, split by regime. Middle 90% of outcomes (5th to 95th percentile):
| Window | Months | 5th | Median | 95th |
|---|---|---|---|---|
| all | 727 | -4.04 | +0.10 | +3.73 |
| 1965-1984 | 240 | -4.55 | +0.38 | +4.58 |
| 1985-2019 | 420 | -2.53 | +0.05 | +2.37 |
| 2020 on | 67 | -4.50 | -0.10 | +5.85 |
The band is about half as wide from 1985 to 2019 as before 1985, and widens again from 2020. So the record does not say “you never know”; it says how much you did not know, in which regime. And over horizons that matter for planning, the real wage:
| Horizon | Independent windows | Negative | 10th pct | Median | 90th pct | Worst |
|---|---|---|---|---|---|---|
| 1 years | ~62 | 40% | -1.6% | +0.4% | +2.1% | -6.1% |
| 5 years | ~12 | 38% | -4.9% | +1.8% | +6.6% | -9.9% |
| 10 years | ~6 | 36% | -9.4% | +4.7% | +9.6% | -13.7% |
| 20 years | ~3 | 40% | -13.6% | +2.0% | +16.4% | -18.9% |
Over ten years the real wage fell in 36% of windows, fell by 9.4% or more in one window in ten, and never by more than 13.7% in this record. Over twenty years it has both fallen (-18.9%) and risen (+20.8%). These are the record’s range for one country and one series, from overlapping windows that are not independent draws (about 6 independent ten-year windows exist), so they are not probabilities to plan on. They do turn “you never know” into “within this range, so far.” Nothing here is investment, insurance or policy advice.
What indexation does to inflation. Whether wages pegged to prices feed back into prices is the central objection to indexation, and this page does not test it. Productivity. Real wages are usually judged against output per hour; not used here. Composition. Who is in the wage series after April 2020 is not computed. An ahead peg. Only the naive trailing forecast is scored. Phillips’s own form. Wage growth against the level of unemployment (Phillips 1958) is still not run; WP-128 §5 ran the price form only. The claim itself. That wages ought to be pegged ahead of inflation is the author’s position, left as such.
Whose wages: a worked case, the teacher who depends on SNAP. For a parent the only real choices are a second job, SNAP, or debt. (The author’s position, left as such. A partner’s income and family help are not counted here. On savings, the Federal Reserve’s 2025 household survey finds 63% of adults could cover a $400 emergency expense in cash and 55% have a three-month rainy-day fund, but only 21% of those with family income under $25,000 and 39% at $25,000 to $49,999 do; CITED, Board of Governors, Economic Well-Being of U.S. Households in 2025.) A teacher entering a one-year residency in New York City is offered a $30,000 stipend, conditional on two years of public-school teaching afterward (Steinhardt, Empire State Teacher Residency Grant; CITED). Spread over 2,080 hours that is $14.42 an hour, and over 1,800 hours $16.67, both under the $17.00 city minimum wage. The MIT Living Wage Calculator for New York County (2026; CITED) gives $38.21 an hour for one adult and $60.92 for one adult with one child, from a required gross income of $126,705, of which housing is $40,877 and child care $16,551. Minutes of work for one gallon of milk, taking the author’s observed $9 and, for comparison, the national average of $4.229 (BLS series APU0000709112 as carried by FRED, August 2026; CITED): at $14.42, 37 and 18; at $17.00, 32 and 15; at $38.21, 14 and 7. The author’s point is that a resident with a family would depend on SNAP, the federal food-assistance program, to buy food at all: $30,000 is $2,500 a month before tax. Whether that falls under the New York gross-income limit for the resident’s household size, and whether the state counts a stipend as income, is not checked; the state agency’s published limits were not retrieved. The $9 price is an observation, not a surveyed figure, and no New York metro gallon price has been found: the two metro series identifiers offered to the author did not resolve to a gallon series. A stipend is not a wage and a peg does not reach it, so the question this case poses is open: whose wages are growing, and does any peg touch the pay of someone entering a high-cost city? Not run here.
Every page tagged ECONOMICS (38, at the commit the script pins), with how many times its own text uses a wage word (wage, real wage, indexation, COLA, cost of living, minimum wage). 2 of them do at all: WP-128, the sibling measure on the same labour-market data, and WP-64. No page in the corpus says “wage–price”, “indexation” or “cost-of-living adjustment” (0 files), so the question this page asks had not been asked here before.
Producing script: book6/wp129-verify.py — recomputes every figure above from FRED (retrieved live), rebuilds the table in section 7 from docs/subjects.tsv at a pinned commit, and checks that this page prints what it computes. CITED FRED series AHETPI, CES0500000003, CPIAUCSL, fred.stlouisfed.org, not held.