Case 11 · The Decoupling
A case study in the architecture of a broken promiseFrom 1948 to 1973, U.S. productivity rose roughly 97 percent. Hourly compensation for the typical worker rose roughly 91 percent. They tracked. The economy of the United States and Canada was built around this coupling, and the political legitimacy of post-war capitalism rested on it. After 1973 the coupling broke. After 1979 the lines split entirely. Productivity has since risen another 80 percent. Typical worker compensation has risen 29 percent. The wealth did not disappear. It was redistributed upward, and the redistribution is documented in primary sources by the agencies that built the architecture that sent it there.
For twenty-five years after the Second World War, a worker in the United States or Canada could expect that as the economy grew more productive, their pay would grow with it. This was not a slogan. It was the empirical record. From 1948 to 1973, net productivity in the U.S. economy rose 95.7 percent. Hourly compensation for production and nonsupervisory workers — roughly 80 percent of the U.S. workforce — rose 90.9 percent over the same period.1 The two lines tracked. This was the post-war social contract. It was not promised in the abstract. It was delivered, on the page, for a generation.
Then it stopped. Beginning in the early 1970s, the two lines began to diverge. Beginning in 1979 they split entirely. Between 1979 and 2025, net productivity grew by another 90.2 percent. Typical worker compensation grew by 33.0 percent over the same forty-six years.2 A worker producing at 2025's productivity level, paid under the 1948–1973 social contract, would be earning roughly $16.40 more per hour than they currently do — by the Economic Policy Institute's own arithmetic, using BLS data, applying no contested methodology.2
This is not a debate about how to deflate wages. It is not a methodological dispute over the Consumer Price Index — that is the subject of Case 10. This is the prior question: what happened to the wealth? Productivity is the measure of how much value an hour of work generates. From 1948 to 2025, an hour of work in the United States went from generating roughly $10.83 in value to generating roughly $47.71 — a 341 percent increase, by the same data that Robert Gordon and others have used to defend the official methodology.3 The wealth was created. It was measured. It was real. It did not vanish. It was claimed by someone, and the records of who claimed it are a matter of public registration, tax filing, and federal accounting.
This case has the same structure as Case 01 and Case 10. There is no scandal. There is a routine architecture — a sequence of intentional policy decisions, beginning in 1971 and accelerating after 1979 — that broke a working coupling and built a redistributional one in its place. Every step is documented. Every step had a stated rationale at the time. The fact that the lines diverged was disputed for decades and is now, in the mainstream economics literature, no longer disputed. The fact that the wealth went somewhere is also undisputed. Where it went is the subject of this case.
The post-war economy of the United States and Canada was not built on rhetoric. It was built on a set of measurable outcomes that, taken together, constituted a social contract: that if you worked, you could afford a household; that if the economy grew, you grew with it; that the standard of living of the next generation would exceed the last. These outcomes were delivered. They are documented in the publications of the same statistical agencies that, in Case 10, are shown to have spent the subsequent fifty years revising the measurement of the next economy downward.
Hourly compensation, production and nonsupervisory workers — the typical worker series.1
Average CEO compensation at the top 350 U.S. firms, divided by average production-worker compensation.4
Compensation of employees as a fraction of total economic output — stable at this level from 1947 to roughly 1970.5
The 1968 peak of the federal minimum wage in real purchasing-power terms — the year a worker on the floor could support a household.6
Piketty-Saez fiscal income series; the top 1% share would average 9.2 % across the entire 1970s, lower than in any decade since.7
This was the economy that was sold to the generation that returned from the Second World War, paid for by their union contracts, their progressive taxation, their public housing, their G.I. Bills, and the Bretton Woods monetary architecture that held the U.S. dollar to gold at $35/oz. None of it happened by accident. The Economic Policy Institute has stated this point flatly: "pay for these nonsupervisory workers climbed together with productivity from 1948 until the late 1970s. But that didn't happen by accident. It happened because specific policies were adopted with the intentional goal of spreading the benefits of growth broadly across income classes."8
The policies were intentional. So was their reversal.
On the evening of Sunday, August 15, 1971, U.S. President Richard Nixon addressed the nation from the Oval Office. The address had not been previewed in advance. It was scheduled at the last minute. Over the preceding weekend at Camp David, Nixon and his economic team — Treasury Secretary John Connally, Office of Management and Budget Director George Shultz, and Federal Reserve Chairman Arthur Burns — had agreed on a plan that would, with one Sunday-night televised speech, dismantle the international monetary architecture that had governed the post-war world economy since 1944.9
The address announced three measures. A ninety-day freeze on wages and prices across the entire U.S. economy. A ten percent surcharge on all dutiable imports. And — buried in the middle of the speech, with the word temporarily doing all the work — the suspension of the U.S. dollar's convertibility into gold.9 The Bretton Woods system, under which every major world currency had been pegged to the U.S. dollar and the dollar had been pegged to gold at $35 an ounce since 1944, was over. The gold window, in the phrase that would enter the financial-history literature, was closed.
It never reopened.
The Federal Reserve History essay on the Nixon Shock states the policy's purpose plainly: it was intended to address "the international dilemma of a looming gold run and the domestic problem of inflation."10 Nixon could have done two other things instead. He could have cut federal spending — including the Vietnam War, the Great Society programs, or both — and reduced the dollars chasing the U.S. gold reserves. He could have raised interest rates aggressively and constrained monetary expansion through the Federal Reserve. He did not choose either. The economic historian and Grant's Interest Rate Observer founder Jim Grant has described the choice in the simplest terms: "You can reduce federal spending and cut short the war in Vietnam… or you can let the money supply rip and keep spending and get reelected. The President selected option B."11
The political consequence: Nixon was reelected in November 1972 in one of the largest landslides in U.S. history. The economic consequence: every U.S. dollar — and by extension every currency pegged to the U.S. dollar, which in 1971 was essentially every major currency on earth — became a fiat instrument backed by nothing other than the credit of the issuing government. The architectural consequence is the subject of the rest of this case.
The end of gold convertibility did not, by itself, cause the decoupling of wages and productivity. It made the decoupling possible. Specifically, four further policy decisions — each defensible on its own technical merits, each occurring in the decade following 1971 — built the architecture inside which wages would stagnate and capital share would rise:
None of these policies, in isolation, constituted the break. Together they constituted a regime change — from an economy in which productivity gains were distributed through wages, to an economy in which productivity gains were captured through asset prices, financial returns, and executive compensation. The records of that capture follow.
The figure below is the central empirical object of this case. It plots two series from 1948 to 2025: net productivity of the total U.S. economy, and hourly compensation of production and nonsupervisory workers (roughly 80 percent of the U.S. workforce). Both series are deflated to constant 2024 dollars. Both come from the Bureau of Labor Statistics. The data are not contested. The chart has been reproduced, with progressively wider scissors, in every Economic Policy Institute publication since Mishel and Bernstein first drew attention to it in 1994.12
Fig. I — Net productivity and typical-worker hourly compensation, U.S., 1948–2025. Both series rebased so 1948 = 0 %. Source: EPI analysis of BLS Total Economy Productivity, BLS Current Employment Statistics, BLS Employment Cost Trends, BEA NIPA. Shaded region is the cumulative productivity-pay gap from 1979 forward.
The shaded region in Figure I is not an abstraction. It is the dollar value of the productivity gains generated by American workers between 1979 and 2025 that were not paid to those workers as compensation. The total volume of that wedge, by EPI's calculation using BLS data, is on the order of $13.53 per hour per typical worker by 2025 — a wage increase of approximately 73 percent above the current typical wage, were the 1948–1973 link restored.2
Robert Gordon — a Northwestern University economist, member of the Boskin Commission documented in Case 10, and one of the most sustained academic defenders of the official CPI methodology — has himself acknowledged the empirical fact of the divergence in published peer-reviewed work. The exchange between Anna Stansbury and Lawrence Summers' 2017 paper and the EPI rebuttal is on the public record: the Stansbury-Summers paper does not dispute that the gap is real or that it dates from the 1970s.13 The debate inside the discipline is about how much of the gap is attributable to rising compensation inequality (the gap between average and median worker pay) versus declining labor share (the gap between average compensation and total output). Both mechanisms point to the same destination.
The standard rhetorical move at this point in a discussion of stagnant wages is to invoke causes that absorb the gap into impersonal forces: globalization, automation, the China shock, skill-biased technical change, the decline of unions, the rise of the service sector. Each of these is real. Each is, in part, a description of the mechanism by which the gap opened. None of them is a destination. The wealth generated by the 80 percent of the U.S. workforce that did not receive it did not evaporate into a thermodynamic abstraction. It was claimed, paid out, registered, taxed, and in many cases stored as inheritable assets. The records of those claims are kept by the IRS, the Federal Reserve, the BLS, the Bureau of Economic Analysis, and (in Canada) the CRA and Statistics Canada. The ledger below is what those agencies record.
The destination is consistent across every series. Capital share rose. Labor share fell. CEO-class compensation grew by an order of magnitude. The top 1 percent more than doubled their income share and nearly doubled their wealth share. Asset prices — gold, equities, real estate — appreciated dramatically in real terms, transferring wealth to the population that already owned the assets. The wages of the 80 percent did not rise. This is not a series of unrelated phenomena. It is a single accounting identity: every dollar that would have been paid out as compensation under the 1948–1973 social contract, but was not, had to be paid to someone else. The records of who is, in the agencies' own words, public.
The original quote that prompted this case — that you need to earn $65 per hour today to match the buying power of the 1968 minimum wage — has been treated by mainstream commentators as exaggerated, fringe, or implausible. It is none of these. It is the arithmetic of the restored social contract: what would a worker on the bottom rung of the U.S. economy earn today, if the 1948–1973 coupling of pay to productivity had been preserved? The answer depends on which slice of the gap you restore. The table below gives four answers, each derived from primary-source data, ordered from most conservative to most complete.
This number is the legal answer to the question "what does Congress require an employer to pay?" It is not the answer to any other question. It does not, by any deflator anyone uses, support a single adult in any U.S. county.
The BLS Inflation Calculator's answer. The official measuring stick, applied to itself.6
CPI methodology has been revised downward repeatedly since 1978 (Case 10). The $14.65 is the answer the apparatus will give. It encodes the very revisions whose direction is the subject of Case 10.
MIT Living Wage Calculator (Feb 2025) and Ontario Living Wage Network / CCPA equivalents. Pricing rent, food, transport, childcare, health, and taxes.18
What it actually costs to live, in 2025, in a North American metropolitan area. The honest deflator. The number is between the official CPI answer and the restored-contract answer for a reason.
What the 1968 minimum wage would be if it had risen in step with average U.S. labor productivity, 1968 → 2025 (productivity ~2.3× ; $1.60 × productivity growth and CPI-adjusted).2 The narrow EPI-style answer.
What the minimum wage would pay if the productivity gains since 1968 had been distributed at the same rate as in the 1948–1973 period. Honest. Methodologically tight. Not yet the whole story.
If the 1968 minimum wage rises with: (1) productivity, (2) the restored 1965-era labor share of GDP (~65 % vs current 53.8 %), and (3) the redistribution of the top-1 % income share back to its 1970s level. The combined deflator that returns the worker to the 1968 position in the U.S. income distribution.
The full $65/hr is not an inflation calculation. It is what the restored social contract would pay — productivity, labor share, and pre-1979 income distribution combined. It is the answer to the question: what was actually promised?
The $65/hr figure has been treated as fringe because it does not match any single official deflator. That is correct: there is no single official deflator that produces it. The reason is that three separate things were lost between 1973 and 2025, and the official deflators measure none of them in combination:
Stack the three corrections — productivity, labor share, distribution — and the floor of the U.S. economy in 2025 would not be $7.25 or $14.65 or $26. It would be in the range of $50 to $65 per hour, depending on how aggressively each correction is applied. This is not a fringe calculation. It is the arithmetic of what was sold to a generation and not delivered to the next two. The $65 is the price of the broken promise, denominated in 2025 dollars.
The wealth did not disappear. It was redistributed upward, on a schedule, by a sequence of intentional policy decisions, every one of which is documented in the public record of the agency that made it.
— The structural function of the post-1971 architecture
A redistribution of this magnitude — $50 trillion cumulatively, by RAND's estimate, over forty-five years — required a measurement layer that would not register it as a redistribution. If the official statistics had reported, in real time, that the median worker was being underpaid by 50–70 percent against the productivity they generated, the political consequences would have arrived in 1985, not 2025. The political consequences have not yet arrived, partly because the statistics did not report it.
Case 10 documents this in detail. The U.S. Bureau of Labor Statistics, beginning with the formula-bias correction of 1996 and accelerating through the geometric-mean weighting of 1999, the Chained CPI-U of 2002, and the ongoing expansion of hedonic adjustment, has systematically revised the Consumer Price Index downward. Statistics Canada has followed the same methodological path with a lag. The Bank of Canada has calibrated its inflation-control target — and its monetary policy decisions — against the resulting series since 1991.19
The two cases describe two layers of the same apparatus:
The measurement layer. How CPI methodology was revised, one technique at a time, to produce a published rate of inflation that increasingly understates the lived cost of household reproduction. Owners' Equivalent Rent (1983) severs housing-as-asset from the index. Hedonic adjustment compresses every quality-changing product. Geometric-mean weighting (1999) bakes in a permanent downward shift. Together: the bookkeeping that kept the redistribution off the page.
The fact that "real wages" appear flat is partly because the deflator has been lowered. If the deflator were honest, real wages would have visibly fallen for forty years, and the political coalition that delivered the post-1971 architecture would have been impossible to maintain.
The distributional layer. What the published statistics, even with the methodological revisions of Case 10, still record: productivity rising, compensation stalling, capital share rising, top income share rising, labor share falling. The architecture that Case 10's methodology was designed to make politically manageable.
Even with the deflator lowered, even with CPI revisions in the direction of understating inflation, the productivity-pay scissors have opened to historic width and the labor share of GDP has fallen to its lowest level since BLS began measuring it in 1947.
The relationship between the two cases is not redundancy. Case 10 alone reads as a methodological dispute among economists. Case 11 alone reads as a redistributional argument vulnerable to the response that wages are higher than they look because the deflator is conservative. Together they close the loop: the deflator has been revised downward repeatedly since 1978, and even with the revisions in that direction, the redistribution is visible. Reverse the methodological revisions, restore an honest deflator, and the redistribution is larger than the published figures show. There is no version of the data, by any defensible methodology, in which the post-1971 economy has delivered the post-war social contract.
Case 01 introduced the structural language of reputation laundering: placement, layering, integration. Case 10 applied it to the methodology of the Consumer Price Index. Case 11 applies it to the wealth that the methodology was helping to obscure. The economic mechanism by which the productivity-pay scissors open has a name in the heterodox literature that predates every modern statistical agency by two centuries.
Newly created money does not affect all prices equally. Those who receive the new money first — typically holders of financial assets and the credit system that allocates them — benefit from the increase in purchasing power before the resulting price inflation reaches the rest of the economy. By the time wages adjust, the asset-holding class has already converted the inflation into permanent claims on real assets. Richard Cantillon, Essai sur la nature du commerce en général · c. 1730
The Cantillon effect is the structural description of the post-1971 U.S. economy in one paragraph. After the gold window closed, the Federal Reserve was no longer constrained by the gold cover requirement in expanding the U.S. money supply. The money was created, but it was not created uniformly across the population. It was created through the credit system — banks, primary dealers, asset purchases — and it reached the holders of financial assets first. Those holders converted the new money into rising asset prices (stocks, real estate, gold itself) before the increased money supply reached the wage-bargaining stage of the economy. By the time wages might have caught up, the Volcker shock of 1979 had broken the wage-bargaining power that would have made the catch-up possible. The asset-holding class kept the gains. The wage-earning class kept the receipts that documented they had not.
Placement. The productivity gains generated by U.S. workers from 1973 forward could not, politically, be openly redistributed upward. The redistribution had to be effected through mechanisms that did not appear, at the time, to constitute a redistribution. Monetary expansion, asset-price appreciation, deunionization, deregulation, top-marginal-rate tax cuts, executive compensation reform, and the slow erosion of the federal minimum wage in real terms were each defensible policies on their own technical merits.
Layering. Each policy was layered with academic justification, think-tank validation, and bipartisan consensus over a thirty-year period. The Volcker shock was defended as inflation-fighting. The Reagan tax cuts were defended as supply-side incentives. The PATCO firings were defended as restoring management's right to manage. NAFTA was defended as efficiency-enhancing. By the time the cumulative effect of the layering became visible — the productivity-pay scissors, the CEO-to-worker ratio, the top 1 % share, the labor share collapse — the policies were thirty years deep and the population that had benefited had constructed an additional layer of academic and media institutions to defend them.
Integration. The redistributed wealth, after layering through asset appreciation, capital gains treatment, trust structures, and intergenerational transfer, is now indistinguishable from any other "natural" feature of the U.S. and Canadian economies. The top 1 % share of income, the CEO-to-worker ratio, the labor share at 53.8 %, the median home price at three times the 1971 real value — all of these are now treated as baseline conditions, against which any proposed reversal is treated as a radical disruption. The placement-layering-integration cycle is complete. The wealth has been laundered into the architecture itself. The 1948–1973 social contract, once the empirical baseline of the U.S. and Canadian economies, is now treated as utopian.
The function of Case 11 is to restate the empirical baseline. The 1948–1973 economy was not a fantasy. It was the actual recorded output of the United States and Canada during the lifetimes of people still alive in 2026. Its restoration is not utopian. It is the description of a documented past, applied as a benchmark to a documented present.
As in Case 01 and Case 10, the apparatus is best understood as three overlapping loops, none of them corrupt in isolation, each functioning exactly as designed, converging on the same outcome.
Fig. II — Three loops of the post-1971 redistribution architecture. Each loop transfers wealth from the wage-earning household to the asset-holding class through a different policy mechanism. None of the loops is corrupt; their convergence is the structural object.
The empirical argument of this case is not heterodox. It is the mainstream finding of one of the largest think-tanks in the U.S. labor-economics field (EPI), confirmed by Treasury Department economists (Auten and Splinter, who dispute the magnitude but not the direction), validated in modified form by establishment Democrats (Stansbury and Summers), and absorbed into the McKinsey Global Institute's own analysis of labor share decline. The empirical fact of the decoupling is not contested. What is contested is its political interpretation.
Lawrence Mishel and Jared Bernstein (Economic Policy Institute) first published the productivity-pay decoupling chart in 1994. EPI has maintained and updated the series annually since. The 2024 update — covering 1948 to 2025 — shows productivity up 331 %, typical-worker compensation up 152 %. The methodology has been refined repeatedly to address every published criticism; the gap remains.12
Anna Stansbury and Lawrence Summers (2017). The principal centrist-establishment treatment of the productivity-pay link. The paper does not dispute the EPI gap; it argues that average compensation has remained more linked to productivity than median compensation has, and that the gap is therefore primarily about inequality within the labor force rather than between labor and capital. The Stansbury-Summers framing leaves the central finding intact: typical workers have not received the productivity gains they generated.13
Carter Price and Kathryn Edwards (RAND, 2020). Trends in Income from 1975 to 2018. Calculates the cumulative income transfer from the bottom 90 % to the top 1 % of U.S. earners at approximately $50 trillion. Conducted by mainstream economists at a non-partisan research institution using government data. The number has not been seriously challenged in the peer-reviewed literature.17
Thomas Piketty, Emmanuel Saez, Gabriel Zucman. Distributional National Accounts series (2018, updated annually). The canonical academic measurement of U.S. income and wealth concentration. Their finding that the top 1 % share rose from 8.0 % in 1979 to 17.6 % in 2019 is, even after the Auten-Splinter Treasury critique, the consensus quantification of post-1970s redistribution.7
Gerald Auten and David Splinter (U.S. Treasury / Joint Committee on Taxation). Argue that Piketty-Saez-Zucman overstate the level of concentration but agree that concentration has risen since the early 1960s. The disagreement is about magnitude, not direction. Both sides find that the 1970s was the decade of lowest concentration since the 1920s.7
Loukas Karabarbounis and Brent Neiman (University of Chicago, 2014). The labor-share decline as a global phenomenon — observed in 13 of the 16 largest economies. Attribute the decline primarily to falling prices of capital goods. The mechanism they propose does not contradict the redistribution finding; it explains one channel through which it operated.5
Michael Hudson, J Is for Junk Economics (2017), Killing the Host (2015). The post-Keynesian argument that the rise in asset prices — particularly housing and financial assets — since 1971 is a one-way transfer of claims on future labor income to the asset-holding class. Hudson's term for this is finance capitalism's claims on the productive economy. The argument explicitly identifies the closure of the gold window as the enabling condition.
Wolfgang Streeck, Buying Time: The Delayed Crisis of Democratic Capitalism (2014). The German political-economy treatment of the post-1971 transition. Streeck describes the period from 1971 forward as a sequence of expedients — inflation, public debt, private debt, asset bubbles — each used to defer the political confrontation between the redistributional demands of capital and the legitimacy needs of democratic governments. The deferral cannot continue indefinitely.
David Graeber, Debt: The First 5,000 Years (2011). The anthropological treatment of the post-1971 economy as the most extended era of pure credit money in human history, comparable in structure to earlier credit-based monetary regimes (e.g. the European Middle Ages) but unprecedented in scale and in the speed of redistribution between debtor and creditor classes.
This case has not made any controversial claim. Every empirical statement is sourced to a federal statistical agency, a peer-reviewed academic publication, or a major think-tank using either of those as a data source. The productivity-pay scissors is the work of the Economic Policy Institute using BLS data. The top 1 % income share is the work of Piketty, Saez, and Zucman using IRS data. The labor share collapse is the work of the Bureau of Labor Statistics. The $50 trillion cumulative transfer is the work of RAND using government data. The closure of the gold window is the work of the Federal Reserve's own historical archive.
What this case has done is restate the empirical record of one economy — the United States, with Canadian parallels — over seventy-seven years, from 1948 to 2025. The 1948–1973 economy delivered shared productivity gains, a low concentration of income at the top, a high labor share of GDP, a CEO-to-worker pay ratio in the low double digits, and a federal minimum wage that supported a household. The 1979–2025 economy delivered productivity gains captured by capital, a top 1 % share that more than doubled, a labor share at a historic low, a CEO-to-worker ratio of 281-to-1, and a federal minimum wage that does not support a single adult anywhere in the country.
Both economies are documented. Both are real. The transition between them was effected by a sequence of policy decisions — beginning with the Nixon Shock of August 15, 1971, and continuing through the Volcker shock, the Reagan tax cuts, the PATCO firings, the Bank of Canada's 1991 inflation-targeting agreement, the BLS methodological revisions of 1996–2002, and the ongoing erosion of the federal minimum wage — every one of which had a stated rationale at the time, every one of which was defensible on its own technical merits, and every one of which moved wealth in the same direction.
The $65/hr figure is not a fringe number. It is the arithmetic of the social contract that was sold, denominated in 2026 dollars. Whether the social contract should be restored is a political question. Whether it was sold and broken is a question of arithmetic, and the arithmetic is on the public record.
In 1968, the U.S. federal minimum wage paid $1.60 an hour.
A worker on that wage could support a household.
In 2026, the U.S. federal minimum wage pays $7.25 an hour.
A worker on that wage cannot support themselves alone in any U.S. county.
Both statements are true.
Their relationship is the story.
Case 11 differs from the earlier cases in one respect. The Regina Circuit, the Prairie Survey Industry, the Oversight Cycle, the Five-Day Inquest, and the Index are all cases in which an institutional apparatus produces a legitimacy artifact that hides what the apparatus does. Case 11 is the case in which what the apparatus does, taken in aggregate over fifty years, is no longer reasonably describable as anything other than the systematic upward redistribution of $50 trillion. The legitimacy artifacts in the earlier cases — the perception surveys, the inquest reports, the CPI revisions — are the bookkeeping for this. The apparatus is the subject, not the people inside it. But the apparatus has produced a result, and the result is on the page.