The April 2026 IMF World Economic Outlook is out. Headline forecasts for 2026:
To make it easier to track how these forecasts evolve, I built the IMF WEO Tracker — a new tool that visualizes and compares WEO vintages going back to 1990, across 196 countries and key indicators. You can see where forecasts converged, where they missed, and how they were revised as new data arrived.
March inflation jumped to 3.3% year over year, up from 2.4% in February. The main driver was gasoline as the Iran war pushed pump prices from $2.98 to $4.15 a gallon. Core inflation held at 2.6%, for now.
The price spike hit already-soft wage growth. Average hourly pay for production and non-supervisory workers rose 3.4% over the year ending March, down from 3.7% in January and February. Adjusted for inflation, real hourly pay grew 0.1% in March, after 1.2% in February. Workers at the bottom, where nominal growth has been slowing the longest, went negative. The 12-month average of real weekly pay at the first decile is down 0.8% over the past year.
Aggregate income tells the same story. Gross labor income, total pay across the workforce, rose 3.7% nominal in March, down from 4.8% a year earlier. Deflated by the CPI, real labor income is growing 0.4%. Real market income, personal income excluding taxes and transfers, is flat, and likely turned slightly negative in March once the CPI spike is reflected.
Three forces are pushing this lower.
First, demographic. A new FEDS Note by Murray and Vidangos shows labor force growth slowing to near zero in 2026, the weakest in modern US history. With immigration collapsed and the population aging, the monthly payroll number needed to hold the employment rate steady has fallen below 10,000. Soft headline prints are partly structural, not just cyclical.
The cyclical side is real, too. Hiring has slowed sharply. JOLTS shows 4.8 million hires in February, down from a 2019 average of 5.8 million. Openings are 6.9 million and the quits rate is 1.9%, near a decade low outside the pandemic. When the labor market runs hot, workers have leverage. They switch jobs, negotiate, and pull wages up. That leverage has faded.
Prices are the third piece, and the March data is where all of this shows up. Before the war, tariffs were already pushing core goods inflation higher. The Iran war then pushed gasoline 39% higher, lifting March CPI to 3.3% year over year. The result is the single worst month for real hourly pay in roughly four years.
Looking forward, the OECD forecasts US inflation at 4.2% in 2026, the highest in the G7. If that path holds, the March flip in real market income will not be a one-month event.
And one cushion households had counted on is gone. Markets priced in a steady path of Fed rate cuts earlier this year, which would have eased mortgage and other borrowing costs. The federal funds target now sits at 3.5 to 3.75%, and those cuts have been pushed out indefinitely. That relief is off the table.
Where people move and where housing gets built tend to go together, and both are concentrated in the US south even as housing prices remain well above their pre-pandemic levels in most areas. Of the 1.8 million people added to the US from July 2024 to July 2025, 1.2 million were in the south, and most new housing supply has been added there as well.
As one result of these changes, California and states in the northeast region will continue to lose congressional seats, and states in the south will add them. In discussions of that apportionment shift, some have pointed out that most of the new housing is also being built there. It only follows that the south will have the population growth. Cause and effect here aren’t clear-cut, but this seems reasonable.
Each month the Census Bureau reports building permits issued in major metro areas. Using these reports, I’ve divided new housing units permitted in each metro area each year by the resident population. The goal is to identify where new housing is likely to grow fastest relative to population. The results are mapped below for the period from January 2024 through January 2026.
Not every permit ends up being built, but the map gives a decent idea of where growth is likely over the coming year. Since the last version of this post, the overall pace of permit issuance has slowed by roughly 10 percent, with nearly all of the decline concentrated in apartment buildings. Higher interest rates have made large apartment projects harder to finance.
The following table shows the top 25 metros by units issued per resident per year. The highest rate is again in the smaller- and mid-sized metro areas of central and southwest Florida, where Wildwood–The Villages, a large retirement community, leads the nation at about 2,440 permits per 100,000 residents per year. The biggest shift from the 2023 version of this post is in Austin, Texas: after leading the large-metro ranks last time, Austin has slipped to 20th. Its early-2020s construction boom appears to have caught up with demand, and the pace of new permits has since eased. Raleigh, North Carolina is now the fastest-building large metro, fitting the pattern: where the south builds, the south grows.
In 1991, Albert Hirschman published The Rhetoric of Reaction, identifying three arguments that have been used to oppose social programs since the French Revolution. He called them the perversity thesis (the program hurts the people it claims to help), the futility thesis (the program accomplishes nothing), and the jeopardy thesis (the program threatens something we already have). All three are in circulation today, and can be understood as stages in a cycle.
The cycle begins with a deliberate fiscal choice. Tax cuts reduce revenue. The resulting deficit becomes the justification for cutting programs that had nothing to do with the shortfall.
This is not a secret. Reagan’s budget director David Stockman told Senator Daniel Patrick Moynihan that “the plan was to have a strategic deficit that would give you an argument for cutting back the programs that weren’t desired.” One of Reagan’s advisers explained to Friedrich Hayek that the president permitted large deficits because they made “absolutely everyone convinced that no more money can be spent.” Grover Norquist put it more plainly: the goal was to shrink government “down to the size where we can drown it in the bathtub.”
The reconciliation bill signed in July 2025 increases the federal deficit by $3 trillion. It also cuts over $1 trillion from Medicaid and SNAP.
With the deficit in place, the jeopardy thesis follows. Programs that were affordable before the tax cut are now portrayed as threats to fiscal stability.
Hayek warned in 1944 that the welfare state would lead to totalitarianism, though he also endorsed a minimum of food, shelter, and clothing for all. Reagan declared in 1981 that “government is not the solution to our problem; government is the problem.” At an Easter lunch in April 2026, President Trump said: “We’re fighting wars. We can’t take care of day care.” The video was posted to and then removed from the White House website. During his 2024 campaign, Trump had promised that tariff revenue would easily cover childcare.
The FY2027 budget requests $1.5 trillion for defense, a 44 percent increase, while cutting domestic programs by 10 percent. Medicaid is put on “a more sustainable path.”
Programs that survive the jeopardy stage are starved of operational capacity. Administrative capacity is cut. Eligibility rules are made more complex. The resulting dysfunction is then cited as evidence that the programs were never effective.
Stuart Butler of the Heritage Foundation described the strategy in the 1980s: build a “parallel system” of private accounts that would gradually create a constituency for dismantling public programs. “You wean people gradually off of social-insurance risk management into private risk management.” The FY2027 budget follows this template. Research grants are replaced by “competitive awards.” The Rural Business Service is eliminated because the Small Business Administration “already” serves rural areas. LIHEAP, which provides heating and cooling assistance to low-income families, is cut on the premise that increased private energy production will make the program unnecessary.
The SNAP cuts in last year’s reconciliation bill illustrate manufactured futility. The bill cuts federal administrative funding in half, expands complex work requirements that increase eligibility errors, and then penalizes states based on their error rates. Arizona has already lost 47 percent of its SNAP participants, including 180,000 children, since the bill took effect.
At the final stage, the cuts are reframed as kindness. Removing people from programs is described as freeing them from dependency.
This is the oldest argument in the cycle. The 1834 New Poor Law in England cut food rations in workhouses and separated families, on the theory that generosity created poverty. Charles Murray recycled the same claim in 1984: “We tried to provide more for the poor and produced more poor instead.” Hirschman noted that except for a slight toning down, the melody was the same, 150 years later.
The perversity thesis has never been limited to one party. Bill Clinton signed the Personal Responsibility and Work Opportunity Act in 1996 to “break the cycle of dependency.” In 2021, Senator Joe Manchin told colleagues that parents would use the expanded Child Tax Credit to buy drugs. The credit, which had cut child poverty to its lowest rate on record, expired. In the current budget, SNAP work requirements are framed as “returned accountability.”
Hirschman observed that people do not gouge out their eyes to qualify for disability benefits. When industrial accident insurance was introduced in Europe, employers claimed workers were mutilating themselves on purpose. The reports were found to be highly exaggerated.
The three arguments require three different versions of reality. Perversity assumes a world so responsive that every policy triggers a backlash. Futility assumes a world so rigid that policy bounces off. Jeopardy assumes the policy is effective enough to be dangerous.
These are opposite claims about how the world works, used to justify the same action: cutting programs. Hirschman observed that this logical incompatibility does not stop people from deploying both arguments against the same policy, sometimes in the same debate.
The arguments are chosen for what they justify, not for whether they are true.
Estimates of last year’s reconciliation bill’s Medicaid and SNAP cuts project a $154 billion reduction in state GDPs, 18 percent more than the $131 billion in federal savings they generate. Households in the bottom 10 percent lose an average of $1,600. Households in the top 10 percent gain $12,000. The arguments rotate. The direction of the transfer stays the same.
The FY2027 budget proposal follows the same pattern. It runs a $2.2 trillion deficit while framing $73 billion in domestic cuts as fiscal necessity. The cuts amount to about 3 percent of the deficit they sit inside.
The next time someone tells you a program costs too much, accomplishes nothing, and hurts the people it serves, all at once, the pattern is worth recognizing. Hirschman’s framework makes it easier to see, and harder to take seriously.
GDP growth and employment growth usually move together. The scatter below plots one-year changes in each, from 1949 to 2025, plus March 2026 in red. The correlation is 0.70.
March 2026 sits in an unusual position: GDP has grown over the past year while employment fell 0.3 percent. That combination shows up a few other times in the data (1958, 1991, 2001, 2009) but in each case the economy was recovering from a recession. There is no recession to recover from in 2026.
A recent FEDS Note by Seth Murray and Ivan Vidangos traces the mechanism. An aging population combined with collapsing net immigration has pushed labor force growth to near zero. The breakeven pace of employment growth has fallen below 10,000 per month, far below any point in the past 65 years. Additionally, federal civilian employment is down 11 percent year-over-year, the steepest peacetime decline on record.
The income side of the economy reflects the job market slowdown. In a healthy economy, US labor income should grow by five percent or more per year. This includes higher wages tied to productivity, cost of living adjustments tied to inflation, and new wages paid to new workers as the workforce grows. Over the past year, amidst a shrinking workforce, nominal gross labor income grew only 3.7 percent. Critically, inflation is now heading toward 3 to 4 percent. The difference between the two measures, real labor income growth, is approaching zero.
GDP growth is decelerating too. After strong quarters in mid-2025 (3.8 and 4.4 percent annualized), Q4 came in at 0.7 percent and the Atlanta Fed Q1 nowcast sits at 1.3 percent. Over the past two quarters, annualized growth averages 1 percent. Coincidentally, Josh Bivens at EPI estimates that about 1 percentage point of recent GDP growth comes from AI-related channels (a stock market wealth effect and capital expenditure) that he characterizes as fragile.
The labor side of the growth equation is stalling. Productivity has to carry everything. That is a lot to ask.
In 1950, the Argentine economist Raul Prebisch published a finding that would shape development economics for decades: the prices of commodities exported by poor countries were falling relative to the manufactured goods they imported from rich ones. The structure of global trade, he argued, transferred the productivity gains of the periphery to the center. Poor countries grew the coffee; rich countries sold the tractors.
Seventy-five years later, something has inverted. China was itself part of Prebisch’s periphery within living memory. Today it is the world’s largest exporter of manufactured goods. And what it exports is solar panels at eight cents per watt, electric vehicles that outsell combustion engines, container ships, and steel. The capital goods that countries need to generate electricity, move goods, and build infrastructure.
The trade data tells this story. The trade data also understates the scale, because trade is measured in dollars, and when prices fall, each dollar carries more physical goods than it used to.
World merchandise trade as a share of global GDP has stagnated since the 2008 financial crisis. Within that plateau, China’s position has transformed.
China is the top import source for 81 countries. The United States is the top source for 29. Germany for 18.
In 1995, China held this position for one country. By 2005, ten. By 2015, fifty-one. China is in the top three import sources for 164 of 207 countries. Only three have China outside their top ten: the Falkland Islands, Gibraltar, and Mauritius.
The countries making this switch span every income level. China is the top source for South Korea, Japan, Australia, and Norway. It is also the top source for most of the world’s lowest-income countries. The median country now sources about 15 percent of its imports from China.
And the pattern of displacement is consistent: countries are switching away from former colonial and hegemonic trade partners. France lost most of francophone West Africa. The United States lost much of Latin America and the Gulf. Japan lost most of East Asia. Russia lost Central Asia. The United Kingdom lost several Commonwealth nations.
What these countries are buying, increasingly, is capital goods.
The conventional image of Chinese exports, cheap consumer goods and textiles, is twenty years out of date. Apparel fell from 8.7 percent of China’s exports in 2005 to 3.9 percent in 2025. What grew was electrical machinery, vehicles, ships, and steel.
The scale of this shift becomes clearer when measured in physical output rather than dollar values. Chinese factories produced 370.9 gigawatts of power generation equipment in 2025, 3.4 times the 2018 level. Motor vehicle production reached 34.8 million units, up from 27.8 million. And output of industrial robots hit 773,000 sets, 5.2 times the 2018 figure. China now installs more industrial robots annually than the rest of the world combined.
These categories reinforce each other. Robots automate the factories that produce vehicles and power equipment. Power equipment runs the factories that produce the robots. The result is a manufacturing system that scales up output while pushing unit costs down, a virtuous cycle that resembles what the Gutenberg press did for books: transforming expensive, labor-intensive goods into mass-market commodities.
When the price of an exported good falls dramatically, each dollar buys more. For the countries importing Chinese capital goods, falling prices amount to a transfer of value from Chinese producers to foreign buyers. Solar panels are a striking example. Using data from Ember, which tracks Chinese customs records on solar exports by destination, we can measure the same trade flow in megawatts of generating capacity, in nominal dollars, and as a share of world GDP.
Since 2018, the physical capacity of Chinese solar exports has increased by more than 700 percent. In nominal dollar terms, exports only doubled. And as a share of world GDP, Chinese solar exports grew by just 59 percent.
The gap between physical and dollar growth reflects a price collapse: the cost per megawatt of exported solar capacity fell about 75 percent. A megawatt of Chinese solar panels that cost $261,000 in 2018 costs roughly $67,000 today.
This price collapse is transforming energy access in the developing world. In Africa, solar panel imports from China surged 60 percent in the twelve months through mid-2025, reaching 15 gigawatts. Sierra Leone alone imported solar capacity equivalent to 61 percent of its 2023 electricity generation. In Pakistan, Chinese panels at eight cents per watt are cheaper than the grid, and the country has avoided an estimated $12 billion in fossil fuel imports since 2021 by deploying rooftop solar. In Cuba, imports of Chinese solar panels grew 34-fold between April 2024 and April 2025, taking the island from zero to roughly 10 percent of electricity generation from solar in about a year. In Lebanon, cheap Chinese solar technology replaced the diesel generator networks that had powered the country for years.
And the price gap is widening beyond solar panels. China’s real effective exchange rate measures, in a single number, how expensive Chinese goods are relative to those of its trading partners, adjusted for inflation. It has fallen 15 percent since early 2022, reaching its weakest level since 2012.
The chart above shows why this matters. The blue line is the nominal exchange rate of the yuan, trade-weighted against China’s partners. It has been roughly stable since 2020. The red line adjusts for relative price levels: it falls when Chinese goods get cheaper relative to the rest of the world. The shaded gap between them has been widening since 2020, driven by diverging price trends. Producer prices in Europe rose 35 percent over the past five years, and 26 percent in the United States. Chinese producer prices over the same period were flat or falling. The same nominal exchange rate now buys considerably more Chinese output than it did five years ago.
The decline is surprising. In standard economic theory, a country with rapid manufacturing productivity growth should see the red line rise. Productivity gains flow to workers as higher wages, higher wages push up domestic prices, and the real exchange rate appreciates. The same pattern held during the industrialization of Japan and Germany. Economists who studied an earlier episode of the same pattern in China called it “China’s real exchange rate puzzle,” and traced it to total factor productivity growth.
The robot data offers a more specific explanation. When productivity gains are embodied in labor, as in postwar Japan, wages rise and domestic prices follow. When productivity gains are embodied in capital, as in factories deploying industrial robots at unprecedented scale, wages face less upward pressure. The cost savings flow into lower goods prices instead. The widening gap between the two lines is the productivity revolution, made visible in prices.
Raul Prebisch observed that manufacturing productivity gains in the United States were, as he put it, “not passed on through prices” but instead absorbed as higher wages and incomes. The prices of manufactured goods held steady while the prices of commodities fell. The peripheral countries that exported coffee and copper saw their terms of trade deteriorate. The rich countries that exported tractors and turbines kept the surplus.
China has inverted this mechanism. Its productivity gains are being passed on through prices. The cost savings from 773,000 industrial robots per year flow outward, through containers of solar panels and electric vehicles, to the countries that import them. At a scale and breadth that has few historical precedents, the terms of trade for capital goods are moving in favor of the buyers.
Falling prices are what countries are responding to. The switch from France or the United States or Japan to China as top import source is, in most cases, a response to prices. Chinese machinery costs less. Chinese solar panels pay for themselves in months. Chinese electric vehicles dominate markets from Thailand to Brazil. The calculus is straightforward, and for importing countries, the outcome is more capital goods for the same money.
Prebisch worried that the periphery would remain trapped, exporting cheap commodities and importing expensive machines. The machines are getting cheaper, and they are arriving without conditions. The countries that need them most are getting more of them than ever before.
Since the Strait of Hormuz closed at the end of February, WTI crude oil has risen from $63 to $105 a barrel. Interest rates across the economy have responded. The bond market is pricing the oil shock as temporary. But the uncertainty it introduced has a longer-term cost.
You can read market expectations for future inflation directly from Treasury bonds. The “breakeven” inflation rate is the difference between a regular Treasury yield and the yield on a Treasury Inflation-Protected Security (TIPS) of the same maturity. It tells you the inflation rate the market is betting on.
Since February 27, the five-year breakeven has risen 15 basis points (hundredths of a percentage point) to 2.55%. The market expects somewhat higher inflation over the next five years, which makes sense given an oil price that nearly doubled. But the five-year, five-year forward rate — a measure of expected inflation in years five through ten — actually fell 3 basis points to 2.07%.
In short: near-term inflation expectations rise; long-term expectations stay anchored. Oil futures tell the same story. The August WTI contract is trading around $79, well below the $105 spot price, meaning the market expects the disruption to fade. That said, the IEA has noted that more than 40 energy sites across nine countries have sustained severe damage, which could extend the disruption beyond what futures currently price.
If markets view the inflation shock as temporary, why have longer-term yields risen? The 10-year Treasury yield is up 33 basis points since February 27. Almost all of that increase, 28 basis points, comes from higher real yields. Inflation expectations barely moved.
The explanation lies in the term premium. When investors buy a 10-year bond, they are locking up their money for a decade. The term premium is the extra compensation they demand for taking on that uncertainty: the risk that the economy, inflation, or policy could go in unexpected directions. The Federal Reserve publishes an estimate using a model called Kim-Wright.
Since the war began, the Kim-Wright 10-year term premium has risen 25 basis points. That accounts for most of the rise in the 10-year yield. Investors are pricing in a more uncertain world. Oil supply disruptions and geopolitical conflict make long-term bonds riskier to hold.
The yield curve plots Treasury yields from the shortest maturities to the longest. Since February 27, it shifted upward across the board, but not evenly. The two-year yield jumped 41 basis points, exceeding the Fed’s overnight rate target of 3.50-3.75%. The 10-year rose 33. The 30-year rose 24.
The short end of the curve reflects expectations about what the Fed will do in the near term. When the two-year yield rises above the overnight rate, it means markets expect the Fed to hold rates where they are. Before the oil shock, core PCE inflation — which strips out food and energy — had already climbed to 3.1% in January, stalling the disinflationary trend from 2023–2024. The oil shock eliminated any remaining room the Fed had to cut.
The Fed is holding rates steady, but monetary conditions are loosening anyway. As headline inflation rises from the oil shock, the real fed funds rate — the policy rate minus inflation — falls without the Fed doing anything. It has already dropped from about 1.0% in February to 0.4% in March. The OECD forecasts US inflation reaching 4.2% this year; if that materializes with the overnight rate unchanged, the real rate would be roughly -0.5%.
This automatic loosening provides a modest cushion on the demand side during the supply shock. But it only works if the Fed holds steady. Raising rates to chase headline inflation would eliminate the cushion and tighten into a supply disruption. This is a shock that monetary policy cannot fix. The supply problem can only be resolved by reopening the Strait of Hormuz or reducing dependence on oil that passes through it.
The unemployment rate has risen to 4.4% — still low by historical standards, but up from 3.4% two years ago. Underneath that number, hiring has stalled, job searches are lasting longer, and a growing number of people have stopped looking altogether. This is what a hiring freeze looks like — and it is a different kind of problem than the recessions our policy tools were built for.
Hiring has slowed steadily.
| 2021 peak | 2018-19 average | Now | |
|---|---|---|---|
| Monthly hires | 6.9M | 5.8M | 5.3M |
| Share of workers quitting | 3.0% | 2.3% | 2.0% |
| Job openings per unemployed person | 2.0 | 1.2 | 0.9 |
Employee turnover hit a nine-year low in January 2026, and the pay bump workers get for switching jobs is at its lowest since 2017. Workers are behaving accordingly: the NY Fed’s Survey of Consumer Expectations shows the share of employed people who expect to voluntarily leave their job fell to 15.9% in February — the lowest in the survey’s history.
The most affected group is people looking for their first job. Unemployment among college graduates aged 22-27 has risen to 5.6%, sharply above the overall rate, with over 40% in jobs outside their field. The Cleveland Fed found that the rate at which unemployed graduates find work has fallen from 47% to 37% since 2000, and that high school graduates now exit unemployment faster than college graduates — a reversal of the historical pattern.
The Sahm rule, a widely watched recession signal based on rising unemployment, triggered in mid-2024 and has since receded. A slow rise in unemployment captures only part of what is happening. The number of people unemployed 15 weeks or longer has risen 45% since 2023. Another 6 million people want a job but have stopped actively searching — a population entirely outside the unemployment rate.
Using the Current Population Survey, I counted everyone who wants work — whether they are actively searching (and counted as unemployed) or have given up searching (and are not counted). As a share of the population, this group has grown steadily since 2023, and totals about 13 million people over the last 12 months of data.
What makes this episode different is where the increase is coming from. In the dot-com recession, the Great Recession, and COVID, the increase was driven mainly by a wave of people losing jobs — newly unemployed workers who had been searching for less than 15 weeks. In the current episode, that group accounts for just 25% of the increase. Nearly half comes from people who have been unemployed 15 weeks or longer, and over a quarter from people who have stopped searching entirely.
Employers treat nine months without work as equivalent to four years of lost experience. Skills atrophy. Job searchers lose motivation. A Minneapolis Fed analysis found that long-term unemployment has trended upward for decades, and that workers who start looking for work during downturns experience earnings losses that persist for years.
This pattern — where time out of work makes it harder to find work — is well-documented among the unemployed. Survey data shows it operates among people outside the labor force too: the longer someone has been wanting work without finding it, the lower their already dismal chances of getting back in — and those chances have been falling.1 This is a population of millions that the unemployment rate does not track.
The labor market entered 2026 with less cushion than it had in 2022. The surplus of job openings that supported the labor market in 2022-2023 has closed, and the pandemic-era savings buffer has been depleted. If hiring remains frozen, people stuck in long job searches and people who have stopped searching have fewer paths back in than they would have had two years ago.
Unemployment insurance was designed in 1935 for temporary layoffs: lose a job, collect benefits, get rehired when demand recovers. That system was built for a different problem. Only about a third of unemployed workers receive unemployment benefits. Young people looking for their first job qualify for nothing at all.
Meanwhile, about 6 to 7 million people per year lose work and want a new job, but never show up in the unemployment statistics because they are not actively searching. Most are part-time or self-employed workers who face severe eligibility barriers. The safety net largely misses them.
There are concrete proposals for reform. One would establish a minimum unemployment benefit available to all jobseekers regardless of work history. Wage-loss insurance — supplementing the pay of workers who find new work at lower wages — would cost an estimated $2.6 to $4.3 billion per year.
Beyond reforming unemployment insurance, some countries invest directly in helping people find new work — through job placement services, retraining, and training programs tied to specific industries. The US spends 0.1% of GDP on these active labor market programs. Denmark spends 2%, and sees substantially higher rates of people finding new work. Similar programs included in US trade adjustment assistance resulted in persistent earnings gains for workers who participated.
Two Democratic senators recently introduced competing proposals to cut taxes for working Americans. Sen. Van Hollen’s Working Americans’ Tax Cut Act would eliminate federal income taxes for individuals earning under $46,000 and couples under $92,000. Sen. Booker’s Keep Your Pay Act goes further, making the first $75,000 in household income tax-free. Both offset part of the cost with higher taxes on the wealthy.
Mike Konczal has described how Democrats arrived here. Bush’s 2001 tax cuts came with expiration dates. When the deadline hit in 2010, Obama — boxed in by a no-tax pledge and 9.8 percent unemployment — extended them. What began as a Republican tax policy became a bipartisan commitment. Eric Levitz pointed out the arithmetic: both senators officially support universal childcare, paid leave, and expanded healthcare. Taxing billionaires cannot fund that agenda and trillions in new tax cuts.
The deeper issue, though, goes beyond fiscal math. It’s what the tax cuts are actually worth. According to the Penn Wharton Budget Model, a family earning $80,000 saves about $1,500 a year under Van Hollen’s plan and roughly $3,700 under Booker’s.
That same family, with employer-sponsored health insurance, pays on average $6,850 per year in premiums for family coverage. Add deductibles and out-of-pocket costs and the total runs $9,000 to $12,000.
If they have a child under five in center-based care, they face an average annual cost of $13,128. And if a parent needs time off for a new child or a family illness, no federal paid leave program exists.
These costs are unavoidable. They are the private price of services that other wealthy countries fund through the tax system. For a middle-income family with a young child, they add up to $20,000 to $25,000 per year.
And yet the United States spends more on these things than other countries, not less. Families just pay out of pocket instead of through taxes.
Among wealthy democracies, the U.S. ranks first in total social spending as a share of GDP. Most of that spending is private — employer health premiums, daycare tuition, out-of-pocket costs. In France, Germany, or Denmark, the same needs are covered publicly. The total cost to society is similar. The difference is who gets the bill.
What would it cost this family to fund these services through the tax system instead?
Healthcare. Under Sen. Sanders’ Medicare for All financing, a 4 percent tax on earnings above $29,000 would cost this family about $2,040 per year. Current premiums and out-of-pocket costs — $9,000 to $12,000 — would be eliminated.
Childcare. The Child Care for Working Families Act would cap costs at 7 percent of family income — a maximum of $5,600, saving about $7,500 compared to the current average. No new household-level tax.
Paid leave. The FAMILY Act would create a social insurance program funded by a 0.4 percent payroll tax split between employer and employee. The worker’s share at $80,000: $160 per year, for 12 weeks of paid leave at 66 percent of wages.
| Costs eliminated | New taxes | Net | |
|---|---|---|---|
| Healthcare | $9,000–$12,000 | $2,040 | +$7,000–$10,000 |
| Childcare (1 child <5) | ~$7,500 | $0 | +$7,500 |
| Paid leave | — | $160 | 12 weeks coverage |
| Total | ~$16,500–$19,500 | ~$2,200 | ~$14,500–$17,500 |
The new taxes total about $42 a week. The private costs they replace total $350 to $400.
There is another gap. The tax cuts leave out the poorest twenty percent of families, who already owe no federal income tax. Universal programs cover them by design. And because middle-class families use the same programs, they stay funded.
The reason matters. Healthcare and childcare are services where private markets fail. Kenneth Arrow established in 1963 that patients cannot evaluate their own care the way they shop for other goods — competitive pricing breaks down. Janet Yellen, as Treasury Secretary, called childcare “a textbook example of a broken market”: unaffordable for parents, underpaying for caregivers, undersupplied for communities.
These are old findings. The reason nothing has changed has little to do with affordability. When you add up what American families already pay in health premiums, daycare, and out-of-pocket costs, they pay comparable rates to Danish or Swedish workers. The difference is that Danish workers get healthcare, childcare, and paid leave in return.
So the question is what to do about it. A tax cut sends families back into the same broken market with the same prices. A program replaces the market. And when government helps families through programs, families know it. When government helps through tax cuts, they don’t. The difference matters: a benefit people can see is a benefit people will fight to keep.
Democrats have spent twenty-five years promising middle-class families they won’t have to pay more in taxes. During that time, private costs for healthcare and childcare have risen relentlessly. By competing on tax cuts, Democrats concede that government has nothing to offer these families. The table above says otherwise.
The government shutdown meant no October CPS was collected. Without it, the BLS couldn’t publish its Usual Weekly Earnings series for the fourth quarter of 2025. That’s a six-month hole in the official wage data.
Using the same CPS microdata the BLS relies on, we can fill the gap. The BD Economics measure tracks the official series closely when both are available (Chart 1). What it shows for the missing period: real median weekly earnings growth fell to -1.0% year-over-year by February 2026.
But here is where it gets interesting. The Atlanta Fed’s Wage Growth Tracker, which follows the same individuals over time rather than taking a snapshot of whoever happens to be working, shows real wage growth of +1.3% over the same period (Chart 2). Workers who kept their jobs are doing fine. The 2.2 percentage point gap between the two measures is not a contradiction. It is a composition effect: the mix of who is working changed.
The Wage Growth Tracker only captures people who report wages in both of their CPS rotation appearances, twelve months apart — and the CPS response rate is now lower than during the pandemic, making that matched sample increasingly selective. If you lose your job between measurements, you disappear from the tracker. That is the point of the design. It filters out composition. But it also means the tracker has nothing to say about the people who left.
We can say something about them. By matching CPS individuals across their rotation groups, we can identify who dropped out of the wage sample and what they were paid before they left. Here is what the dropout profile looks like compared to the 2023-2024 baseline:
| 2023-2024 average | 2025 Q4 | |
|---|---|---|
| Survey Retention rate | 66.5% | 60.3% |
| Dropout median wage | $1,072 | $1,146 |
| Stayer median wage | $1,256 | $1,296 |
| Federal dropout wage | $1,386 | $1,716 |
The share of wage workers still reporting pay twelve months later fell to 60%, matching COVID-era lows. The median wage of the people who left jumped to $1,146 per week, roughly $75 above baseline. Federal workers who left were paid $1,716 per week. The people leaving the wage sample are not marginal workers. They sit above the median.
The composition effect in wage data usually runs one direction: in a downturn, lower-paid workers lose hours or jobs first, and the remaining sample looks artificially richer. This time it is running the other way. Higher-paid workers are leaving, pulling the cross-sectional median down while individual wage growth holds steady. The wage data looks worse than individual experience because of who is being cut.
This is the statistical footprint of a specific policy. About 330,000 people left the federal workforce in 2025 while only 110,000 were hired. Workers with 30 or more years of experience saw a 33% net decrease. The most experienced, highest-paid public workers left at the highest rates. Meanwhile, scientists are leaving the country, European Research Council applications from US-based researchers nearly tripled, and the Pentagon cut ties with MIT and Carnegie Mellon in favor of Liberty University and Hillsdale.
Wage composition effects are normally a measurement nuisance that economists adjust for. This one is different. It is not cyclical churn that will reverse when demand recovers. The positions were abolished — not for fraud or waste, but ideology. The expertise walked out the door. The wage data registered it as a composition shift. The country will register it as 180-minute Social Security callback times, unfilled nuclear security posts, and research labs relocating to Aachen and Marseille.
The official data went dark for six months. What it missed was not a recession. It was a restructuring of who employs skilled workers and on what terms. The composition data is the receipt.
BD Economics Blog 