How much damage has █████ done to the working class?

The Pattern

The first three years looked better than the ending. Unemployment fell to 3.5 percent by February 2020. Real median household income rose, poverty fell, and workers at the lower end of the labor market finally gained some bargaining power. Then the pandemic arrived, and about twenty-two million payroll jobs disappeared across March and April 2020 (Bureau of Labor Statistics). A presidency that had claimed the expansion inherited the collapse.

That sequence makes the question difficult. A president is not the economy. He does not personally set every wage, create every job, or cause every recession. If the question is how much damage one administration did to the working class, the honest method is to isolate policies whose effects can be measured: tariffs, taxes, labor standards, health and nutrition benefits, and now artificial intelligence.

On those measures, the record is not ambiguous enough to be comforting. The administration repeatedly placed risk on people who live from wages while concentrating gains among people who live from capital. Tariffs made households pay more. Tax cuts produced a small wage return beside a large corporate benefit. The 2025 reconciliation law shifted resources away from the poorest households. Even its technology policy follows the same rule: owners get speed, scale, and discretion; workers get exposure, retraining, and a lecture about adapting.

█████

Start with the evidence that cuts the other way

The pre-pandemic labor market was strong. In February 2020, the unemployment rate was 3.5 percent, and private production and nonsupervisory workers earned an average of $23.96 an hour. Their inflation-adjusted hourly earnings had increased 0.7 percent over the preceding year, according to the Bureau of Labor Statistics. The Census Bureau reported that real median household income increased 6.8 percent in 2019 under the methodology then in use, while the official poverty rate fell to 10.5 percent.

Workers benefited from that labor market. The statistics do not prove that the president created it. The expansion began in 2009, unemployment had been falling for years, and Federal Reserve policy, congressional spending, demographics, and inherited momentum all mattered. The fair conclusion is narrower: workers gained ground during the first three years, and the administration governed during that gain.

The pandemic is an equally poor shortcut. By February 2021, ten million people were unemployed, but a global disease caused much of the shock. Blaming every lost job on one officeholder would turn chronology into causation. The firmer case begins where researchers can compare products, firms, industries, or households exposed to a particular policy with those that were not.

█████

The tariff bill came home

The first-term trade war is the cleanest example. Economists Mary Amiti, Stephen Redding, and David Weinstein found that the 2018 tariffs passed almost completely into the domestic prices of imported goods. Foreign exporters did not absorb the charge as advertised. American importers and consumers did. By the end of 2018, the researchers estimated a reduction in aggregate US real income of $1.4 billion per month in the Journal of Economic Perspectives.

A separate peer-reviewed study by Pablo Fajgelbaum, Pinelopi Goldberg, Patrick Kennedy, and Amit Khandelwal estimated a $7.2 billion aggregate real-income loss after accounting for tariff revenue and gains to protected producers. The figures measure different periods and concepts, so they should not be added together. They point in the same direction: the policy transferred money from purchasers to the government and selected producers while leaving the country poorer overall (Quarterly Journal of Economics).

The promised manufacturing payoff did not survive closer inspection. Federal Reserve economists found that industries more exposed to the tariffs experienced relative employment declines. Protection helped some producers, but higher input costs and retaliation outweighed those gains. Producer prices rose, and manufacturing output did not (Federal Reserve study).

The second-term tariffs repeated the price mechanism. Federal Reserve researchers estimate that tariff changes through November 2025 raised core-goods prices by 3.1 percent through February 2026 and lifted the overall core personal-consumption-expenditures price level by 0.8 percent. They found that pass-through was effectively complete by then (Federal Reserve). CBO’s earlier estimate for the tariff schedule in place between January 6 and May 13, 2025 projected that it would add an average 0.4 percentage points to inflation in 2025 and 2026 while reducing real output. That projection is not a measure of every later tariff change, especially after subsequent litigation altered the policy, but it identifies the same burden: reduced purchasing power (CBO).

█████

Capital received the certainty. Workers received the forecast.

The 2017 tax law initially reduced federal liabilities for most households. CBO documented that fact, and an honest accounting has to keep it. The harder question is what workers received for the law’s large corporate provisions.

Research using administrative tax data found that the business tax changes increased investment. They also reduced corporate tax revenue by about 40 percent relative to its previous level. The authors estimated that long-run GDP rose by less than 1 percent and labor income by less than $1,000 per employee, far below the administration’s prediction of $4,000 to $9,000 (peer-reviewed assessment; underlying research). The result was not zero. It was a weak labor return on an expensive and immediate benefit to capital.

Public Law 119-21 made the distribution more visible. CBO projects that, between 2026 and 2034, the law will reduce resources for households in the lowest income tenth by about $1,200 per year, or 3.1 percent of their income. Households in the highest tenth gain about $13,600 per year. Losses at the bottom come mainly from reduced Medicaid and Supplemental Nutrition Assistance Program benefits (CBO distributional analysis).

CBO also projects that the law’s Medicaid changes will increase the number of people without health insurance by 7.5 million in 2034 (2026 economic outlook). These are projections, not injuries already completed. They describe enacted policy, not a campaign promise.

█████

One more risk transferred downward

Artificial intelligence is not the center of this record. It is another place where the governing instinct becomes impossible to miss. Executive Order 14110 had directed agencies to study displacement, workplace surveillance, labor standards, and support for affected workers. Labor Department principles called for consultation, transparency, protection of labor rights, and better job quality (official text; Labor Department principles). The framework was incomplete and much of it was voluntary. It at least treated workers as people whose consent and security mattered before deployment.

That order was revoked in January 2025. Executive Order 14179 told agencies to identify and remove barriers to AI innovation, and the resulting AI Action Plan centered acceleration, infrastructure, adoption, and technological dominance (Federal Register). It offers apprenticeships, education, measurement, and rapid retraining. It does not give workers a general right to know when automation will replace positions, bargain over deployment, or share the productivity savings. Owners are invited to move faster. Workers are invited to become employable again.

The employment evidence does not justify an AI apocalypse story. Danish research found no measurable effect on earnings or recorded hours in the first two years of chatbot adoption (NBER). But Stanford researchers found a 16 percent relative employment decline among workers aged 22 to 25 in the most AI-exposed occupations, and BLS projects customer-service employment will fall by 153,700 positions from 2024 to 2034 while software-development employment grows (Stanford Digital Economy Lab; BLS). The honest conclusion is narrow: the national collapse has not happened, but some young workers are already losing the first rung of the ladder.

The damage is a distribution

The administration governed during real pre-pandemic gains. It also imposed tariffs whose costs were paid largely at home, delivered a small worker return beside a large corporate tax benefit, enacted future cuts concentrated among poor households, and treated worker security as something to discuss after owners had been given what they wanted.

There is no honest grand total. The units do not combine. Tariff welfare losses, foregone overtime, lost health coverage, benefit reductions, and entry-level jobs never created cannot be placed in one column without pretending that unlike things are interchangeable.

But refusing a fake total does not require moral confusion. The pattern is visible. The cost arrives as a higher grocery bill, a benefit removed, an insurance card lost, or an opening that disappears before anyone can apply. The gain arrives as protected revenue, a lower corporate tax bill, a higher valuation, or a productivity increase owned by someone else. One side is told the pain is temporary, necessary, or character-building. The other side receives the money.

That is not an accident of presentation. It is the policy. The working class is asked to finance protectionism at the register, subsidize capital through the tax code, absorb benefit cuts in the name of fiscal discipline, and meet technological displacement with personal reinvention. Then the people collecting the gains call the arrangement prosperity.

The damage is not one spectacular collapse. It is a transfer repeated until insecurity feels like weather, followed by the demand that workers be grateful it is not raining harder.