The productivity-pay gap

Economic Policy Institute: net productivity of the US economy has grown ~160% since 1973, while real hourly compensation for non-supervisory workers has grown ~20%. The gap, which did not exist 1948-1973, is the single largest long-run economic fact about US labor. It is the arithmetic of the extraction thesis: workers produced the additional output; capital kept the additional value.

The federal minimum wage has been $7.25/hour since July 24, 2009. Adjusted for inflation, that purchasing power has fallen ~28% over the same period. It is the longest stretch without an increase in the history of the minimum wage (enacted 1938). 20 states still use the federal minimum as their ceiling; the other 30 states + DC set their own higher floors. A full-time minimum-wage worker earns ~$15,080/yr before taxes.

Economic Policy Institute meta-analysis of state enforcement data: US employers steal roughly $50 billion per year from workers via unpaid overtime, off-the-clock work, tip pocketing, misclassification, and minimum-wage violations. The federal Wage and Hour Division recovers ~$30 million per year, or 0.06%. Wage theft exceeds the combined total of all other property crime in the US by a factor of ~2×. It is the single largest category of crime by dollar value; it is rarely prosecuted.

Piketty, Saez, Zucman: real pre-tax income for the bottom 50% of US earners was essentially flat from 1973 to 2022 when measured across decades, rising only when tax transfers are added. The top 1%'s real income roughly tripled over the same period. The bottom-half line is the median US labor experience of the last 50 years: productivity growth, inflation, and economic expansion have gone past, not through.

BLS: 9.9% of US workers belonged to a union in 2024, down from ~33% in 1955. Private-sector rate: 5.9%. Public-sector rate: 32.2%. The private-sector figure is the lowest since unions were legalized federally in 1935. Declines in the 1980s-90s were policy-driven (PATCO firings 1981, NLRB appointments, right-to-work laws); the 2000s-2020s decline is structural (service-sector shift, misclassification, employer legal tactics).

Gallup August 2024: 71% of Americans approve of labor unions, the highest reading since 1965 and up from a 2009 low of 48%. Approval crosses partisan lines: 88% of Democrats, 71% of independents, 50% of Republicans. Support has been climbing for 15 years but organizing rates have lagged; the structural barriers to certifying a union (employer anti-union campaigns, card-check absent, limited penalties) are the practical gap.

National Labor Relations Board: union election petitions rose 52% between FY2021 and FY2023, the highest filing rate in 25 years. Starbucks, Amazon, REI, Apple retail, and graduate-student campaigns drove the spike. Unfair labor practice charges against employers rose in parallel; the NLRB has a persistent backlog and limited remedies even when employers are found to have illegally fired organizers.

EPI: US employers hire specialized "union avoidance" firms during 75%+ of organizing drives. Per-worker spending runs from $5,000 to $50,000 depending on firm size and campaign length. Captive-audience meetings (mandatory anti-union sessions) are legal in federal law; they are banned in most peer OECD countries. The economic incentive to break a campaign is strong because the NLRA penalty for illegally firing an organizer is typically reinstatement plus back pay, which most dismissed workers never pursue.

FTC: ~30 million US workers (roughly 20% of the workforce) are bound by non-compete agreements that restrict where they can work after leaving a job. The FTC issued a nationwide ban on non-competes in April 2024; a federal district court blocked the rule in August 2024, and the ban has not taken effect. Non-competes disproportionately affect low- and middle-wage workers (fast-food workers, hairdressers, janitors); the original justification was trade-secret protection for executives.

EPI: approximately 56 million US private-sector workers (~60%) have signed employment contracts requiring them to resolve disputes via private arbitration rather than court, and 41% of those clauses also bar class-action claims. Arbitration outcomes favor employers approximately 80% of the time; average damages are ~1/3 of court awards. The Supreme Court's Epic Systems ruling (2018) confirmed mandatory arbitration clauses are enforceable against class-action waivers.

The US is the only OECD country and one of only four countries in the world without a national paid family leave policy. The FMLA (1993) guarantees 12 weeks of unpaid leave, but only to employees at firms with 50+ workers who have been there 12+ months, which excludes ~44% of the workforce. 13 states + DC have their own paid programs; most offer 6-12 weeks at 60-80% of wages. Peer-country averages: 15-52 weeks at full pay.

Gallup: ~16% of US workers earn primary income through independent-contractor, gig, or platform arrangements. These workers are excluded from federal minimum-wage, overtime, unemployment insurance, and employer-side payroll-tax contributions by classification. DOL's 2024 final rule tightened the test for misclassification; courts have narrowed it on appeal. Platform companies (Uber, Lyft, DoorDash, Instacart) funded the 2020 Prop 22 campaign in California ($205 million, a state record) to preserve contractor status.

Autor, Dorn, Hanson (AER 2016): the 'China shock' of 1999-2011 caused a net loss of ~2 million US manufacturing jobs, concentrated in specific counties that have not recovered. The paper upended the trade-economics consensus that adjustment costs would be short-run and localized losses would be absorbed. Subsequent work (Autor et al. 2020) tied the same counties to deaths of despair and political realignment. The China-shock literature is the empirical origin of the bipartisan protectionism revival 2017-present.

Goldman Sachs / OpenAI 2023-24 analysis: approximately 30% of US occupations are estimated to have tasks substantially exposed to large-language-model automation. Legal, finance, software engineering, and administrative support rank highest. A separate Anthropic / Brookings 2025 study found early wage-premium compression in roles where LLM use was already pervasive. The aggregate labor impact is projected, not yet realized; early entry-level white-collar hiring has weakened in 2024-25 in the exposed roles.

BLS final 2025 data shows the US added 181,000 nonfarm jobs over the year, averaging ~15,000 per month, the weakest calendar year since the 2020 pandemic. Initial monthly reports had been ~48,000/month; a March 2025 benchmark revision alone stripped 911,000 phantom jobs. December 2025 unemployment rose to 4.4%, partly because discouraged workers left the labor force; average duration of unemployment hit 24.4 weeks, the longest since 2021. This is the labor-market consequence of the tariff cycle.

Economic Policy Institute: S&P 500 CEO-to-median-worker pay ratio was ~20:1 in 1965 and ~290:1 in 2024. CEO compensation rose ~1,200% from 1978 to 2023 (inflation-adjusted); typical-worker compensation rose ~15% over the same period. The gap has widened in every decade since 1980 and is not explained by firm size, skill, or productivity changes at the CEO level. It is the compensation-side equivalent of the productivity-pay gap in the first section of this page.