Two rooms, one question: are they following you, or just obeying you?

I’ve spent thirty years leading people in manufacturing plants, and I ask myself one question every time I evaluate my own leadership: are people following me because I’m in charge, or because I’m taking them somewhere better? If the answer is the former, I’ve told myself the job isn’t leadership anymore. It’s just supervision with a title.

The difference shows up fast on a plant floor. People who follow because you’re in charge do exactly what you tell them, exactly when you’re watching. The minute you leave the room, the work slows down, corners get cut, and nothing moves unless someone’s standing over it. People who follow because you’re taking them somewhere better do the opposite. They keep going when you’re not there. They fix problems you didn’t ask them to fix. They tell you the truth when something’s wrong instead of waiting to get caught. One of those groups is complying. The other one is invested.

I’ve hired and fired based on that difference more times than I can count. A manager who can command a room but can’t get anyone to move without direct orders isn’t leading. They’re just occupying the position.

I think about that question differently now when I think about how we vote.

For at least the last fifteen years, an honest look at our national politics shows the same pattern repeating on both sides. Campaigns built almost entirely around beating the other side. Entire platforms defined by what a candidate is against rather than what they’re actually going to build. Elections framed as existential battles where the other party winning is treated as the real disaster, regardless of what either side is actually proposing to do about the cost of living, the job market, or anything people deal with at their own kitchen table.

That’s the same trap as a manager who leads through fear of consequences instead of through where the team is headed. It works, in the sense that people show up. It doesn’t work in the sense that matters, because nothing is actually moving anyone toward a better place. It’s just two sides making sure the other side doesn’t get there first.

Here’s the test I’d ask anyone to run on their own voting record, regardless of party: over the last fifteen years, when you voted for someone, were you voting for what they were going to build, or were you voting against the person on the other side? Be honest about the answer, because most of us, myself included, would have to admit it’s been more of the second than the first.

That’s not a failure of any one party. It’s a pattern both parties have leaned into because it works. Fear of the other side turning out to vote is a more reliable motivator than a hard sell on a policy platform, and campaigns follow what motivates people. If beating the other side keeps winning elections, beating the other side is what we’ll keep getting offered.

The plant floor version of this problem has a straightforward fix. You stop leading through fear of what happens if people stop obeying, and you start leading by showing people where you’re actually taking them. The country’s version is harder, because it means voters have to stop rewarding candidates who win by making the other side look like the disaster, and start asking what any given candidate is actually going to build.

I don’t know which candidates pass that test right now. That’s genuinely not the point of this. The point is the test itself, because right now, most of us aren’t applying it. We’re voting the way you’d follow a boss who never once told you where the company was headed, just kept telling you why the other guy running it would be worse.

So ask yourself the same question I ask myself. Over the last fifteen years, has your vote been about getting somewhere better, or about making sure the other side didn’t get there first? Those are not the same question, and only one of them builds anything.

Debate the argument with me.

Want the next one delivered instead of stumbled on? Subscribe at christinemarletti.substack.com.

About the Author

Christine Marletti is a Marine veteran, operations executive, and Six Sigma Black Belt with thirty years of plant operations leadership across 7 industries and 4 countries, and the author of The Alignment Economy (July 2026). Read more at christinemarletti.substack.com or the-alignment-economy.com.

The gap between what work pays and what life costs is being bridged with borrowed money

At the grocery checkout, a quiet kind of math happens. A woman who works full time, who has never missed a shift, watches the total climb and runs the same calculation as everyone in line behind her: what’s in the cart against what’s in the account. Then she reaches for the card — not the debit card, the other one — because the distance between those two numbers has to be covered somehow, and this is how. She isn’t reckless. She isn’t living beyond her means. She is living exactly at her means, and her means no longer reach the end of the month.

The most recent reading, for June, actually cooled: prices rose 3.5 percent over the past year, down from 4.2 percent the month before, and wages grew almost exactly as fast. For one month, the gap that usually shows up at the register nearly closed. It closed for a specific reason worth knowing: energy prices fell sharply, and by the time the report came out, oil prices were already climbing again on renewed conflict overseas. Economists reading the report flagged the relief as likely temporary, not a change in direction. A single calmer month doesn’t undo what four decades of an uneven pace built, and it doesn’t change what the woman at the register has been living through to get here.

That near-parity is new, and it’s worth taking at face value rather than either dismissing it or leaning on it too hard. It came from a sharp, energy-driven drop in the headline number, not from paychecks suddenly growing faster. The report itself was volatile underneath the surface — the categories that fell hardest were the ones that swing the most, month to month, for reasons that have nothing to do with wages. One month of relief sitting on top of four decades of drift is not the same thing as the drift ending.

And this is not one strange month. The gap the woman is covering has been widening for a long time. For roughly four decades, what a person can earn has drifted away from what a life costs — slowly, unevenly, but in one direction. Output per worker has climbed steadily across those years; the typical worker’s pay has trailed far behind it. Each year’s slippage was small enough to absorb. Stacked end to end, the small slippages became the thing she is standing in at the register.

So how does a country whose wages don’t cover its costs keep spending? It borrows. American household debt has reached a record 18.8 trillion dollars. Credit card balances sit near their all-time high, around 1.25 trillion, up almost six percent in a single year and more than 300 billion dollars above where they peaked before the pandemic — carried, on average, at an interest rate north of 21 percent. When you hear that the American consumer is resilient, that spending is holding up, this is a large part of what’s holding it up. Not rising paychecks. Rising balances. The strength is partly borrowed, and the loan charges interest.

It isn’t spread evenly. The borrowing, and the falling behind, concentrate exactly where you’d guess — among the households whose wages fall furthest short of their costs. Economists have started calling it a K-shaped consumer: one branch climbing, one branch sinking, sharing a single misleading average. The people covering the widest gap are the ones with the least to cover it with, and the delinquency numbers are beginning to show the strain. A bridge loan only works until the bridge has to be repaid.

Step back, and the distance between a paycheck and a life gets filled three ways. The first is debt — the card as a bridge over the end of the month. The second is public assistance — the other bridge over the same gap. And here a familiar instinct enters: if people need help, they should work to earn it. It sounds like plain common sense. But look at what the record actually shows. The adults these work requirements target overwhelmingly already work, or qualify for an exemption. When Arkansas became the first state to attach a work requirement to its health coverage, employment did not rise — but roughly 18,000 people lost that coverage, most of them not because they refused to work but because they couldn’t complete the paperwork proving they already did. Federal budget analysts have noted that few of those dropped would pick up coverage through a job instead. Requiring work as the price of support moves a person from dependence to independence only if the work itself can carry them. When it can’t, you haven’t ended the subsidy. You’ve changed whose ledger it sits on — from a public one to a private one, made of debt, or of simply going without.

That third patch is the quietest, and the one no debate counts: the care postponed, the prescription stretched, the appointment skipped, the thing not bought. It never appears in a spending report, because it is the absence of spending. It’s the gap left open, carried silently by the household itself.

You can argue about which patch you prefer — more credit, more assistance, more quiet stoicism from people already doing everything asked of them. But the argument over patches skips the only question that matters. A market needs workers who can afford to live on what the work pays. We have built one in which, more and more, they can’t — and then we act surprised when the balances climb.

None of this requires a villain, and reaching for one gets the problem wrong. No single decision arranged it. Prices rose for their own reasons; wages lagged for theirs; the safety net and the credit card filled in where the paycheck stopped reaching. Each step was ordinary. Together they produced an economy that produces plenty and increasingly cannot sell it back to the people who make it, except on credit.

The woman at the checkout will get through the line. She’ll cover the gap the way she covered it last month, and she’ll go to work tomorrow, because none of this is a failure of effort on her part. The demand is real. The work is real. The willingness was never in question. The only thing missing is the connection between what her work produces and what her life costs.

This is the gap the National Productivity Unit, or NPU, is built to close. It ties a wage floor to two things at once: half to what the economy produces per person, half to the actual cost of the basics a household can’t skip — food, shelter, utilities, getting to work. Right now nothing ties a paycheck to that second number directly. Prices move, and a wage has no built-in reason to move with them. The difference gets filled by one of the three patches this piece describes — a card at 21 percent interest, a program with its own paperwork burden, or a family simply going without.

A floor built from the real cost of living closes that gap before it opens, instead of leaving it to be covered after the fact. NPUX does the same work at the level of one worker’s own output — pay that moves when what a person produces moves, the same year it moves, not absorbed silently for four decades and then discovered at a credit card statement.

The woman at the register needed her paycheck to reach the end of the month. Now there’s a design for a part that makes sure it does.

Want the next one delivered instead of stumbled on? Subscribe at christinemarletti.substack.com.

In 2009, one of my employees asked me to hold his annual raise. He was earning $18.00 an hour. A 2.5% increase — $450 a year — was going to cost him nearly $600 a month in lost benefits.

He was not gaming the system. He was reading it correctly. And that is the problem.

For fifty years, serious people from both parties have looked at the American economy, identified something real that was wrong, and applied a serious correction. Some of those corrections came from the right — lower taxes on capital, reduced barriers to business formation, stronger returns to shareholders. Some came from the left — higher wage floors, work incentives, benefit expansions, earnings credits for working families. Each intervention was designed carefully. Each addressed a genuine symptom. And the underlying problem persisted through all of them.

This is not an argument that the policies failed. Most of them worked exactly as designed. It is an argument about what working as designed actually produced — and why the problem kept moving rather than getting smaller.

The pattern has a name.

Think of it as squeezing a balloon. Apply pressure at one point and the air moves somewhere else. The balloon does not deflate. It changes shape. Every major economic policy intervention of the past fifty years has squeezed the balloon in some form. The pressure moved. The shape changed. The same underlying problem resurfaced somewhere new.

That pattern is not a coincidence. It is the economic equivalent of an old medical joke: a patient goes to the doctor complaining that everywhere he touches himself, it hurts. After extensive testing the doctor says he cannot find anything wrong. Show me where the pain is, he says. The patient presses his finger to his shoulder. Winces. Presses his finger to his knee. Winces again. After half a dozen locations the doctor calls the nurse in and says: put this man’s broken finger in a splint.

Fifty years of economic policy. One broken finger. A lot of very expensive examinations of the wrong locations.

From the right.

The 2017 Tax Cuts and Jobs Act reduced the corporate rate from 35% to 21% — the largest single corporate rate reduction in American history. The theory was explicit: lower the cost of capital, investment increases, hiring increases, wages rise. Treasury projected average household income would increase by $4,000 annually.

Capital responded rationally. Fixed business investment rose 9.4% in 2018. S&P 500 companies executed approximately $806 billion in share buybacks that same year — a record, up 55% from 2017. The $4,000 household income projection did not materialize on the predicted timeline. Wage growth in the two years following the cut was not meaningfully different from wage growth in the two years before it, when the labor market was already tightening.

The capital was freed. It went where capital rationally goes when the connection between corporate performance and worker compensation has already quietly loosened. The policy worked. What it assumed — that freed capital would flow back to workers — was already broken before the legislation passed.

From the left.

Dozens of states raised minimum wages over the past decade. The research is genuinely mixed — which is itself informative. In strong labor markets with large employers, wage floors held reasonably well. In weaker markets, something consistent happened: hours fell, benefits eroded, automation accelerated, prices adjusted. The floor moved up. The ground shifted underneath it. Take-home gains were reliably smaller than the nominal wage increase predicted.

In 2018, Arkansas implemented Medicaid work requirements for able-bodied adults. The theory: connect benefit receipt to workforce participation, reduce dependency, increase employment. Within seven months, approximately 18,000 people lost coverage. A Harvard study found the majority were already working, in school, or exempt — they met the requirement but could not navigate a monthly online reporting system. Employment rates in the affected population did not increase. A federal court blocked the program, finding it failed its own stated purpose.

The requirement worked as designed. The design was wrong about what the problem was. The doctor treated the knee. The finger stayed broken.

The best case we have.

The Earned Income Tax Credit is broadly considered the most successful anti-poverty program of the modern era. Ronald Reagan called it the best anti-poverty program ever devised. It has genuine bipartisan support. It rewards work, which is exactly what it was designed to do.

And in the income band where working families are trying to build stability — roughly $19,000 to $50,000 for a single parent with two children — the EITC phase-out stacks on top of federal income tax, payroll tax, SNAP phase-outs, housing assistance cutoffs, and childcare subsidy cliffs. The Congressional Budget Office has confirmed that effective marginal tax rates for households in this range can exceed those paid by the highest earners in the country.

Nobody designed that trap. Serious people designed each program independently, each solving a different symptom. The trap assembled itself from the interaction of the solutions.

The thing every intervention assumed.

Every policy described above shared one assumption: that the connection between what the economy produces, what workers earn, and what it costs to live inside that economy was basically intact — and that applying targeted pressure at one point would correct the visible imbalance.

That assumption was wrong. The connection began drifting in the early 1970s. Since 1948, productivity has grown approximately 290%, while compensation for the median worker grew roughly 115%. Labor’s share of national income hit a historic low of 53.7% in the first quarter of 2026, the lowest level recorded since the government began tracking it in 1947. Every intervention on this list was, in some form, a response to symptoms of that drift.

None of them closed it. Not because the efforts were wrong. Because none of them were aimed at the right thing.

What was missing was not a better patch. It was a built-in connection — a single, baseline economic index designed to hold what the economy produces, what workers earn, and what participation costs in a defined, automatic relationship. Not a tax. Not a floor. Not a requirement. A connection, built in rather than bolted on, that makes the transmission automatic rather than dependent on policy getting it right from outside.

That connection now has a name and a design: the National Productivity Unit. Half its formula tracks national output per person — the productivity side of the fifty-year drift above. Half tracks the actual cost of covering the basics — the side every patch in this piece was really trying to reach, and kept missing from a different angle. The two are tied together automatically, the same way Social Security already ties benefits to inflation, instead of waiting on the next targeted intervention to guess correctly from outside.

The part was never built. Fifty years of data shows what happens when it isn’t. Now there’s a design for what happens when it is.

Want the next one delivered instead of stumbled on? Subscribe at christinemarletti.substack.com.

In 2021, American households had something they hadn’t had in a long time.

A cushion.

The pandemic transfer payments, the paused student loans, the months of nowhere to spend money — all of it stacked up into something that looked, briefly, like financial stability. The personal savings rate hit 26.6% in March 2020 and stayed elevated through 2021. For the first time in a generation, a significant share of working households had a buffer between themselves and the next emergency.

That buffer is gone.

By June 2026, the personal savings rate stood at 2.7% — essentially unchanged from April’s 2.6% and May’s 3.0%. That’s not a single bad month. That’s three months running at a level the country has seen only once before, briefly, in 2005. The rate sits at less than one-third of the long-run historic norm of 8.4% — the average American household saved at that rate for the entire postwar period from 1959 through the early 2000s. It held at 5.4% through 2024. It fell to 4.6% in 2025. And it was nearly half of what it had been just twelve months earlier, in April 2025 — a 49% year-over-year decline in the share of income households were able to set aside.

The sand in the hourglass isn’t falling. It’s evaporating.

What the Historic Norm Actually Means

The 8.4% figure isn’t an abstract benchmark. It represents what American households, across decades and across income levels, were able to preserve out of every dollar they earned after taxes.

At 8.4%, a household earning $60,000 a year was saving roughly $5,000. Not aggressively. Not heroically. Just routinely — the way you’d expect a person to set something aside when the math between what they earn and what they spend leaves room for it.

At 2.7%, that same household is saving about $1,620. Before a car repair. Before a medical bill. Before a week of missed work. The buffer that once absorbed the routine shocks of a working life has been reduced to something that disappears with the first disruption.

There’s a reason the number landed here, and it’s sitting in the name of the statistic itself. The personal savings rate measures saving as a share of “disposable” income — income left over after taxes. It says nothing about what’s left after the actual cost of staying alive. For a full-time minimum-wage worker, the bare-survival basket — a room to live in, food, utilities, transit, one phone, nothing else — has run between 106% and 168% of take-home pay since 1997. Even a $15-an-hour floor, twice the federal minimum, still leaves survival costs eating 87% of take-home. “Disposable” income, for most of the workforce, is a line in an accounting table. It was never a real cushion to begin with.

Disposable income and survivable income have never been the same number for most of the workforce.

The current reading sits well below anything the series recorded outside that one single month in 2005 — not from a crisis, not from a recession, but from the accumulated weight of a cost structure that has been outrunning household income for years.

Where the Cushion Went

The narrative that gets attached to declining savings is usually some version of consumer recklessness. People spending on things they don’t need. Lifestyle creep. The failure of individual discipline.

The data doesn’t support that story.

What the data shows is a household sector that ran down its pandemic reserves to cover a cost of living that didn’t slow down when the transfers stopped. Essential costs — shelter, food, healthcare, transportation, utilities, a phone — have risen approximately 190% since 1979. Worker productivity, the actual output that funds the economy, rose 90% in the same period. The cost of staying in the game grew roughly twice as fast as the economy’s ability to pay for it.

The pandemic savings were not a windfall that got spent frivolously. They were a temporary pressure relief — a valve that opened for two years while costs continued climbing and wages continued their decades-long failure to keep pace. When the valve closed, the pressure didn’t go away. It went somewhere else.

That somewhere else is what a companion piece on this site, “The Strong Consumer Is Running on Credit,” lays out in full.

How the Gap Got Built

Before 1979, something held the floor in place. Productivity grew at 2.5% a year. Compensation grew at 2.1%. They weren’t identical, but they were close enough that the average working household could expect the cost of a basic life to stay roughly within reach of what they earned.

After 1979, the rates diverged. Productivity kept growing — slower, at 1.4% a year, but still growing. Compensation slowed to 0.6% a year. That gap, compounded over forty-five years, is 57 cents of every productivity dollar that didn’t reach a paycheck.

For a long time, the gap was papered over. Savings eroded gradually. Consumer credit expanded. Two-income households became the norm where one had been enough. The adjustments were slow enough that no single year looked catastrophic.

The pandemic savings were the last piece of paper. And now they’re gone.

What 2.7% Actually Means

A savings rate of 2.7% means that for every $100 a household earns after taxes, $2.70 goes to savings and $97.30 goes to everything else.

And borrowing against it is exactly what’s happening. Total household debt hit $18.8 trillion in Q1 2026 — an all-time record. Credit card balances stand at $1.25 trillion, $325 billion above their pre-pandemic record. That’s not the next article’s data getting ahead of itself. That’s the destination the evaporating hourglass was always pointing toward.

The cushion is gone. The debt is what replaced it.

That’s not a spending story. That’s a math story.

What Comes Next

The savings rate decline is the first domino. When households run out of reserves, they don’t stop needing things. They finance them. And the cost of financing — at 21.52% average APR on revolving credit — turns a temporary shortfall into a permanent condition.

The pressure cooker didn’t explode. It just ran out of steam. And now something else is under pressure.

This is the gap the National Productivity Unit, or NPU, is built to close. Right now nothing connects a household’s income to the actual cost of covering the basics, so whatever room a family has left to save is just whatever happens to survive the race between the two. A floor tied half to national output and half to the real cost of the basics doesn’t guarantee a fat savings account. It stops the floor itself from eroding under a family that’s doing everything right.

The cushion didn’t disappear because people stopped being careful with it. It disappeared because nothing was holding the floor still while the cost of standing on it kept climbing. An hourglass with no sand left can’t be fixed by asking people to be more careful with sand that isn’t there anymore. It can be fixed by making sure the household has some to begin with.

Want the next one delivered instead of stumbled on? Subscribe at christinemarletti.substack.com.

Every few months the government releases a number that’s supposed to be good news for everyone. The most recent one, covering early 2026, was another strong showing: productivity rose 2.9 percent from a year earlier, ahead of the long-run average the country has posted since the government started keeping score in 1947. Productivity is booming again, and the headlines cheered.

For most of the last century, that number meant something simple for ordinary people. When the line ran faster, the paycheck got bigger. Rising productivity and rising wages moved together, and the engine that drove them was manufacturing. A factory that made more per hour could pay more per hour, and it did — lifting millions of people without a college degree into a stable, middle-class life. That was the deal, and for decades it held.

I spent thirty years running industrial operations, so I watched that engine up close. And I watched it stop.

Manufacturing productivity didn’t keep climbing. Since 2010 it has gone backward — output per hour in the sector actually shrank, after rising more than three percent a year for the two decades before. The engine that once turned efficiency into broad prosperity quit producing the efficiency. And when economists looked closely, they found the old gains had never been as broad as they seemed: nearly all of manufacturing’s productivity growth came from a handful of computer-related industries. The rest of the floor was carried.

So, if productivity is back, where is it coming from now? Not the factory. It comes from a narrow band of the economy — information and software, finance, professional services — and within those, from a thin slice of firms. The most productive companies, the top five percent, saw their productivity rise more than half in manufacturing and two-thirds in services over a recent stretch, while everyone else gained almost nothing. The same split shows up by worker: the industries posting the biggest gains are the ones with the most degrees and the most artificial intelligence in the building. The growth is real. It’s just standing in a very small number of places.

Here is the part that turns a statistic into a wage story. Those high-productivity firms are built differently. They produce enormous output with relatively few workers — which means that as they come to dominate their industries, the share of all that output going to wages falls. Not because any single company cut anyone’s pay, but because the economy’s center of gravity shifted toward firms that pay out little of what they make.

The same report that clocked productivity at 2.9 percent this spring carried its quiet twin in the very next line: labor’s share of everything the economy produced fell to 53.7 percent, the lowest level recorded since the government started tracking it in 1947. Not a separate claim from a separate study. The same release, the same quarter, two numbers pulling in opposite directions.

The Federal Reserve has traced why separately: as industries concentrate, labor's share of income drops on its own — and a falling labor share means economic growth stops turning into raises.

That is the whole misalignment, traced to its source. We’ve spent years puzzling over how the country can grow richer while ordinary paychecks stall. This is how. It isn’t that productivity died. It’s that productivity moved — out of the broad, many-handed places that shared it, and into narrow, few-handed ones that don’t. The gap between what the economy produces and what workers earn isn’t a mystery, and it isn’t a moral failing. It’s a matter of geography: where the growth happens, and who’s standing there when it does.

None of this is an argument against progress. The new productivity is a genuine achievement, and the answer is not to slow it down. The answer is to notice what the country noticed a century ago, when the assembly line first made a few men rich and then — once wages caught up to it — made a nation prosperous. Productivity lifts everyone only when something connects it back to the people who aren’t in the room where it happens. Manufacturing used to be that connection, by accident of its structure. Nothing has replaced it.

So, when the next productivity number lands and the headlines cheer, it’s worth asking the second question, the one the number hides: more output, yes — but reaching whom? The booming part of the economy can keep right on booming and leave most of the country exactly where it stands. That isn’t a flaw we have to accept. It’s a missing connection — and connections can be built.

This is the gap the National Productivity Unit, or NPU, is built to close. Half its formula is tied to total national output per person — the same broad productivity figure that just came in at 2.9 percent. The other half is tied to the actual cost of covering the basics. When output rises anywhere in the economy, the floor is designed to rise with it, whether the gain showed up on a factory floor or inside a software firm nobody outside the industry has heard of. It doesn’t ask which five percent of companies produced the gain. It only asks whether the number that measures the whole economy’s output moved — and if it did, the floor moves too.

Manufacturing used to make that connection by accident, because paying more per hour was how a factory kept its best workers. Nothing built that connection into the industries where the accident doesn’t happen naturally. A floor tied to the whole economy’s output does deliberately what manufacturing used to do by chance.

Want the next one delivered instead of stumbled on? Subscribe at christinemarletti.substack.com.

My father drove a truck for 42 years. He believed something simple: work hard, show up, and the math works out. For most of his career, it did. A truck, a house, a family raised on one income, and something left over at the end.

His grandson works just as hard, with more schooling than his grandfather ever had. The math doesn’t work out the same way. Rent alone can eat what a paycheck used to cover for rent, groceries, and a car payment together.

Ask Americans if hard work still leads to getting ahead, and the answer has been sliding for a quarter century. In 2001, Gallup found 76 percent of Americans believed it did. The following year that number hit a record 77 percent. By 2014 it had dropped to 54 percent. In January of this year, it stood at 49 percent.

That single reading moves with the political weather, since belief about the country’s direction tracks who’s in office. So it’s worth reading alongside a steadier number. Every January since 2001, Gallup has also asked a broader satisfaction question and averaged the results for the year. That average opened at 48 percent in 2001. This year it fell to 36 percent, the lowest reading in the 25 years Gallup has tracked it. That’s not a swing tied to one election. It’s a quarter-century decline that continued through four administrations and both parties holding power.

Something changed. It wasn’t the work ethic.

Since 1979, the amount each American worker produces has grown by 90 percent. Take-home pay for a typical worker, adjusted for inflation, has grown by 33 percent over that same stretch. Economists calculate what that gap means in a single paycheck: the typical worker earns about $16.40 less per hour today than they would if pay had simply kept pace with what they were producing. That’s not a rounding error. That’s a mortgage payment.

The generational wealth numbers show where that gap landed. Baby Boomers, who spent most of their working years on the near side of 1979, now hold just over half of all household wealth in the country. Generation X holds about a quarter. Millennials and Gen Z combined, despite now making up most of the working-age population, hold about one-ninth. That isn't a story about ambition. It's a story about which generation was working while pay still tracked output, and which generation has spent its whole career after that connection came apart.

The easy move here is to argue about which decade or which party caused it. That argument has run for twenty-five years and produced nothing but the numbers above. The more useful question is why the country still doesn’t have a plain instrument that shows, in real time, whether a family’s paycheck is keeping pace with what it actually costs to hold a normal life together: food, rent, a car that runs, the basics of staying employed. We track inflation. We track the stock market. We track GDP. None of those numbers point at that question, which is exactly why every dial can read fine while a family quietly falls behind.

There’s a plain number that could do this: call it the National Productivity Unit, one that ties pay and the cost of participating in the economy to actual productivity growth, the same way Social Security already adjusts for inflation. If the country gets more productive, staying in the workforce shouldn’t quietly get harder to afford. Right now it does, and nothing in the system is built to notice.

My father never asked for a formula. He asked a simpler question: if I work hard and do the job right, will the money follow? For 42 years, mostly, it did. His grandson is asking the same question today, doing the job just as right, and the honest answer is: not the way it used to. Twenty-five years of declining satisfaction is what happens when a country keeps asking that question without ever building the instrument to answer it.

Want the next one delivered instead of stumbled on? Subscribe at christinemarletti.substack.com.

U.S. fertility rate, 1960–2026. Altitude = children per woman. The flame beneath each balloon represents the purchasing power available to act on the desire for a family. Sources: CDC/NCHS; Gallup.

Gallup has been asking Americans the same question since 1936: what is the ideal number of children for a family to have?

The answer has barely moved in fifty years. Americans consistently say 2.7. More than 80% say the ideal family has at least two children. The number has held through recessions, through social upheaval, through every shift in culture that commentators use to explain why young people aren’t having children.

The birth rate has not held. It sits at 1.6 — a historic low, well below the 2.1 needed to replace the population, and 1.1 children below what Americans say they want.

That gap doesn’t appear in the data because Americans changed their minds about families. It appears because the costs required to build one left the paycheck behind.

Demographers have spent considerable energy explaining the decline as a cultural shift — delayed adulthood, changed priorities, a generation that wants different things. The surveys don’t support that reading.

When researchers ask people who are not having children why, the answer is money. A 2024 Pew Research study found that 36% of adults under 50 who are unlikely to have children cite affordability as a major reason. The Population Reference Bureau finds the same pattern: the primary driver is a growing gap between the family sizes people say they want and the families they feel they can afford.

The timing of the decline makes the cause clear.

Between 1980 and 2007, the U.S. birth rate fluctuated within a stable, narrow range. It didn’t track political cycles. It didn’t track cultural movements. It held.

Then 2007 happened. The birth rate fell sharply and immediately, tracking the economic collapse. That much is familiar — births drop in recessions. What didn’t happen is the recovery. As the economy came back, the birth rate didn’t. A May 2026 working paper from the Federal Reserve Bank of Atlanta documents the pattern: instead of rebounding, birth rates continued their steady decline through the 2010s and into the 2020s, reaching the current historic low of 1.6.

The recession didn’t cause the decline. It revealed a cost structure that had quietly made family formation unaffordable for a growing share of working Americans — and when the 2008 shock forced the calculation into the open, people did the math and the math didn’t work.

Consider what the math looks like today.

The national average annual cost of center-based infant care is $15,636, according to the 2025 Child Care Aware of America Price and Supply report. In California it reaches $22,628 — more than double the average in-state public college tuition of $11,314 in that state. Nationally, two children in center-based care cost $28,191 per year. That figure exceeds median rent in 49 states and the District of Columbia. It exceeds the average mortgage payment in 45 states.

This bill arrives at the beginning of a career, when earnings are at their lowest. There is no loan for it. No 30-year amortization. No insurance pool spreading the cost across a lifetime. It is due monthly, from whatever is left after rent and healthcare and the student debt that was required to get the job that pays for all of it.

The costs that matter for family formation — housing, childcare, healthcare — have one thing in common with each other and one thing different from everything else.

What they have in common: all three have risen sharply in real terms since the late 1970s.

What makes them different from other costs: you cannot outsource them, substitute them, or wait for technology to make them cheaper. A family needs shelter. A working household with children needs childcare. The price of both is set by local markets with no international competition and no productivity gains to pass along.

Meanwhile, real wages for the majority of American workers have been roughly flat since 1979. The Economic Policy Institute has tracked the divergence for decades: productivity and pay moved together from 1948 through the late 1970s, then split. Productivity kept climbing. Pay didn’t.

The policy conversation around birth rates tends toward incentives. Baby bonuses. Tax credits. Direct payments for new parents. Some countries have tried these at scale. The evidence suggests they produce modest, temporary upticks at best.

The reason is straightforward. A tax credit pays a family a little for a child. It does nothing about what a single paycheck can buy, which is the thing that changed. You cannot subsidize your way back to a life the paycheck used to cover on its own.

The 1.1-child gap between what Americans say they want and what they have is not a mystery. It is not a generation that decided children weren’t worth the trouble. It is a generation doing arithmetic on a cost structure that wasn’t designed with them in mind.

The birth rate is a receipt. It records what one ordinary income — or two — can still reach. That number has been shrinking for fifty years. The receipt doesn’t lie.

If you want to stay ahead of pieces like this one, subscribing is the easiest way — christinemarletti.substack.com.

Net productivity growth vs. typical worker pay growth, 1979–2025. Sources: EPI Productivity-Pay Tracker, BLS, BEA.

Picture a drivetrain. An engine on one side, turning at full power. A gear on the other side, meant to receive that power and carry it somewhere useful. Between them, teeth that are supposed to mesh.

When a gear strips, the teeth shear down or wear smooth. The engine doesn’t slow. It keeps turning at full speed, full output. But the gear on the receiving end starts to slip. Less of that power makes it through with every rotation, until eventually the engine spins freely and the gear barely moves at all. Nothing exploded. Nothing broke all at once. The teeth just quietly stopped catching, a little more each cycle, until the two sides that used to move together stopped moving together at all.

That’s the picture behind the numbers. Since 1979, worker productivity in the United States has grown 90.2 percent. Typical worker pay, adjusted for inflation, has grown 33 percent. One side of the drivetrain is turning at full power. The other side caught barely a third of it.

Before 1979, the two numbers move almost as one line. From 1948 to 1979, productivity grew about 2.5 percent a year and pay grew about 2.1 percent a year, close enough that a rising economy meant a rising paycheck for most people doing the work. After 1979, the annual numbers split. Productivity growth slowed some, to about 1.4 percent a year. Pay growth slowed more, to about 0.6 percent a year, less than a third of the earlier rate. A small gap compounding quietly every year for four and a half decades is how you get a gear that looks intact from across the room and is stripped down to bare metal up close.

There’s a second metric that shows the same slippage from a different angle. Labor’s share of national income, the portion of everything the economy produces that actually reaches the people producing it, held near 63 percent for most of the postwar period. It has since fallen to 53.8 percent, the lowest reading since the government started tracking it in 1947. Roughly nine of those points disappeared just since 2000. Two different metrics, measuring two different things, both showing the same gear losing its grip.

Put a dollar figure on it and the picture gets sharper still. If pay had kept pace with productivity since 1979, the typical worker would be earning about $16.40 more per hour today. Economists estimate that of every dollar of productivity gained since then, 57 cents never made it back to the worker who helped produce it. That’s not a rounding error. That’s most of a dollar, redirected somewhere else, year after year, at scale, across the entire workforce.

Here’s the part worth sitting with. Nobody designed the gear to strip. For a couple of postwar decades, pay tracked output closely enough that it looked automatic, and it was easy to assume the connection would just hold on its own. It held because of habit and bargaining leverage that happened to be strong at the time, not because anything was built into the system to keep it that way. When those informal forces faded, there was no part underneath doing that job instead. The gear had never actually been cut to mesh permanently. It just hadn’t been tested yet.

That’s not a story about greed or laziness, and it’s not a story that splits along party lines. A drivetrain doesn’t care who’s driving. It either has the part that keeps power and output connected, or it doesn’t. This one never did. The slippage since 1979 is what forty-five years of running without that part looks like on paper.

This is the gap the National Productivity Unit is built to close, and it does it in two parts. The NPU itself sets the floor: half from national output per person, half from what it actually costs to cover the basics of participating in the economy. That’s the baseline gear, cut to mesh with both sides of the equation instead of drifting loose from either one. NPUX is the second part, the one that captures what an individual worker actually contributes, their role, skill, and output, as a position above that floor. Right now the system has neither piece. Pay floats free of national output on one side and free of individual contribution on the other. NPU and NPUX are the two gears cut to catch both.

The engine was doing its job the whole time. So was the person standing next to it, watching the gear spin loose and wondering why the machine felt like it was working against them. What was missing were the parts that keep pay meshed with output, at the national level and the individual one. Now there’s a design for both.

If you want to stay ahead of pieces like this one, subscribing is the easiest way — christinemarletti.substack.com.

A household's need for an ordinary thing has a ceiling. Past it, more money stops turning into more buying

A family of four needs about four bicycles. One for each person. Give that family a good year — a raise, a bonus, a windfall — and they might buy nicer bikes, lighter frames, a fifth one for the garage. What they will not do is buy forty. A household’s need for any ordinary thing has a ceiling, and past that ceiling more money does not turn into more buying. It turns into something else.

Hold onto that, because it explains more about the economy than it first appears to.

Demand is not unlimited. It is capped inside each household and spread across all of them. The country’s appetite for bicycles — or refrigerators, or haircuts, or restaurant dinners — is the sum of millions of households each buying roughly what they need. Which means demand is a creature of breadth. It is largest when income is wide, when many families each have enough to buy their four. It shrinks when income narrows.

This is where the arithmetic turns, and it turns against the intuition most of us carry. We tend to assume that money moved to the top still gets spent, just by different people. But a need with a ceiling does not scale with a bank balance. Hand one household a thousand times the income of its neighbors, and it still buys about four bicycles. The buying that nine hundred ordinary households would have done — that demand does not move up the ladder with the money. It simply never happens. The dollars are there. The purchases are not.

I want to be precise about what this is and is not, because the precise version is the stronger one. This is not a complaint about wealth, or a claim that anyone behaved badly. No one cheated. It is a mechanical property of ordinary goods: a household’s need for them runs out of room to grow. Run a rising share of a nation’s income into households that have already bought what they need, and you are pouring water into glasses that are already full. The water doesn’t disappear. It just stops doing the one thing demand is supposed to do, which is get spent.

The standard reply at this point is that concentrated income isn’t lost — it becomes investment, and investment builds the factories and funds the jobs. That used to be truer than it is now. Increasingly, income that pools at the top doesn’t go to building new capacity that hires people and fills shelves. It goes to bidding up the price of assets the top already owns.

Money that flows into the price of existing stock and existing real estate is not a paycheck and not a customer. It is a number getting larger. It does not come back down the chain as demand.

And here is the part that ought to bother people across every political line. The households whose unfilled buying would keep the production lines running are, very often, the same households whose work made the production possible. Output per worker has roughly doubled across the last two generations. The pay of the typical worker has not come close to following. The people on the floor produced the surplus. They would have spent it on the things the surplus became. Instead the money routed past them at both ends — it didn’t reach them as wages, and so it couldn’t return as sales.

None of this is a new idea. It is old capitalist common sense, and some of its clearest spokesmen were not reformers but industrialists and bankers. In 1914, Henry Ford doubled his workers’ wages to five dollars a day. He was not being generous, and he said so plainly. A worker who can afford the car is a customer for the car. Ford understood that he was not just employing the people on his line — he was manufacturing his own market, one paycheck at a time.

Marriner Eccles understood the other half. Eccles was a conservative Utah banker, no one’s idea of a firebrand, who went on to chair the Federal Reserve under Franklin Roosevelt. He described the economy before the Crash as a poker game. As the night wears on, the chips drift to a few good players. The rest stay in the game only by borrowing, and for a while the borrowing keeps the table alive and everyone pretends nothing is wrong. But when the losers’ credit finally runs out, the game stops — not because anyone broke the rules, but because there is no one left with chips to bet. The table goes quiet with most of the money still sitting on it.

That is the position an economy like this one drifts toward when income concentrates: a table where one or two players hold the chips and the rest can no longer ante. It can run a long time on borrowing — we have proven that. It cannot run forever.

The fix is not to take the chips. It is to keep the game going — to make sure the people producing the value can afford the things the value produces. That is not charity. It is the condition a market needs in order to clear. We have built an economy extraordinarily good at producing and steadily worse at putting purchasing power where the producing happens.

The demand never left. It is still sitting in every household in the country, capped and patient, four bicycles at a time. The only thing missing is the income to meet it — and that is a thing a system can be built to fix, once we are willing to call it what it is. Not a fight over who deserves what. A broken connection between making and earning, waiting to be reconnected.

This is the gap the National Productivity Unit, or NPU, is built to close. It sets a floor that moves with what the economy actually produces — half tied to national output per person, half tied to the real cost of covering the basics. As the country gets better at making things, the floor rises with it, instead of leaving that gain to pool wherever the org chart happens to route it.

Put that back inside the bicycle household. Right now, when output rises, the extra value has no guaranteed path back to the people who produced it — it can just as easily end up bidding up the price of an asset someone else already owns. A floor tied to output closes that gap directly. It doesn’t hand anyone a fortune. It makes sure the four-bicycle household actually has the income to buy its four, every time the economy gets bigger — which is the exact purchase Eccles’ poker table needed and never got. NPUX, the second part of the system, does the same job at the level of one worker’s output instead of the whole economy’s — pay that moves when what a person actually produces moves, not pay that sits still while productivity climbs somewhere above it.

Ford didn’t need a theory to see this. He needed customers who could afford the product coming off his own line. A hundred years later, the country still hasn’t built the part that does that automatically. Now there’s a design for one.

Want the next one delivered instead of stumbled on? Subscribe at christinemarletti.substack.com.

AI adoption is arriving in neighborhoods that have no say in how fast it moves or what it costs them

Oracle filed its annual report with the SEC on June 22, 2026. In it, the company disclosed that it had reduced its workforce by 21,000 people, about 13% of its total headcount. The stated cause, in a legally binding document, was this: the adoption and deployment of AI technologies across our operations.

Not restructuring. Not market conditions. AI. Named directly. In a 10-K.

That is a different category of evidence than a think tank projection or a CEO interview. Companies lie to journalists. They do not lie to the SEC.

The productivity gains from AI are real. That part is not in dispute.

Goldman Sachs estimated in 2024 that generative AI could raise annual U.S. labor productivity growth by roughly 1.5 percentage points. The Federal Reserve has begun tracking AI-adjacent sectors and finding wage premiums for workers whose jobs complement the technology. A 2024 NBER working paper by MIT economist David Autor found that AI has the potential to extend decision-making capability to workers who currently lack access to it. This, if the technology is deployed in that direction, could rebuild some of what computerization hollowed out over the past four decades.

The gains are real. The question is where they go.

The technology doesn't ask the neighborhood's permission before it lands

Because the pattern so far looks familiar.

Automation hits tasks first, not jobs. That is what a 2025 NBER paper from researchers at Yale, Northwestern, and MIT found after tracking over 58 million LinkedIn profiles and 14 million job postings from 2014 to 2023. Tasks with higher AI exposure saw reduced demand for the skills attached to them. But the workers in those jobs did not simply disappear. They reallocated effort toward tasks the AI had not yet reached.

The net employment effect was modest. The productivity effect at AI-adopting firms was not. One standard deviation increase in AI adoption correlated with 12% higher productivity, 31% higher sales, and 42% higher profits.

The firms that adopted AI grew faster. The workers in those firms saw some benefit. The workers at firms that did not adopt AI fell behind, not because their jobs were automated, but because their employers grew more slowly.

That is the pattern the data keep finding.

The U.S. economy has run this play before. Computerization from the 1980s onward did something similar: it concentrated decision-making power in elite experts, the minority of workers with college or graduate degrees, while automating away the middle-skill jobs that once formed the backbone of the working class. Clerical work. Administrative support. Blue-collar production. The workers displaced from those jobs did not move up. They moved into lower-paid service work that computers could not do because it required physical presence.

AI is different from computerization in one important way. Computerization complemented workers at the top. AI, according to the research, is hitting the tasks that those same workers perform: the diagnosis, the document production, the code review, the legal brief. The exposure is concentrated near the 90th percentile of the wage distribution, not the middle.

That is new. Whether it helps or hurts people in the middle depends entirely on what happens next.

Autor’s argument is that it could help. AI capable of performing some of what doctors, lawyers, and engineers do could extend the reach of those capabilities to nurses, paralegals, and technicians who currently have the foundational training but not the institutional access. He calls this the missing opportunity of the current moment.

He also calls it a choice, not an inevitability.

The data from the 2014–2023 period, before generative AI arrived at scale, show the gains concentrated in firms that adopted AI. Those firms were already larger, more productive, and higher paying before they adopted. The technology made the gap bigger.

A 2026 NBER working paper that surveyed academic economists, AI industry professionals, and expert forecasters found broad agreement that the direction of change is toward greater concentration. Under a scenario where AI advances rapidly, economists forecast that the top 10% of households would hold 80% of total national wealth by 2050, up from 71% today. Blue-collar occupations would fall to 8% of the labor force from 13% now. The labor share of total economic output, already at its lowest point since records began in 1947, would fall further.

None of that is certain. But that is where the weight of the evidence points.

The connection between what the economy produces and what workers take home has been slipping since roughly 1979. GDP per capita has more than doubled in real terms since then. Real median household income has grown about 35%. The gap between those two lines is not an accident. It is the result of a system that was never built to keep them together.

AI does not create that problem. It scales it, in both directions at once. A worker can suddenly produce far more with AI’s help, and the system has no part that moves their pay when that happens. The gain goes wherever the org chart decides it goes, the same way it has since 1979.

This is the gap the National Productivity Unit and NPUX are built to close. NPU anchors the floor nationally, half from GDP per capita, half from what it actually costs to participate in the economy, so the floor itself can’t be quietly scaled away from workers the way the labor share has been for four decades. NPUX is what would actually answer Autor’s question. If AI genuinely extends a paralegal’s or a nurse’s capability closer to what a lawyer or a doctor produces, NPUX is the part that lets their pay position move with that expanded output, instead of freezing in place while the software vendor and the employer split the difference. Right now, extending capability and extending pay are two separate events, and only the first one is guaranteed to happen.

The productivity is real. Whether it reaches the people doing the work is a design question, and for the first time there’s a specific design on the table that answers it.

Want the next one delivered instead of stumbled on? Subscribe at christinemarletti.substack.com.

Two divers are working the same stretch of ocean floor. One is carrying a large tank. One is carrying a small one. Neither gets a say in the depth, and neither gets a say in what happens next: a fixed station on the seafloor takes exactly two liters of oxygen from every diver who passes, no matter how much they’re carrying.

The diver with the large tank barely notices. Two liters is a small fraction of a big supply. The diver with the small tank notices immediately — the same two liters is a large fraction of a much smaller supply. Nobody adjusted the toll for either diver. The toll was never the problem. The math around it was.

That’s the shape of a flat-fee tax. It doesn’t ask what a household has left after paying it. It just takes the same number, and lets income do the rest of the work — or fail to.

Start with the one that’s been in the news: tariffs.

A tariff is a charge on a good coming into the country. It isn’t scaled to the buyer’s income — a household earning $30,000 a year pays the same tariff on a pair of shoes as a household earning $300,000. The Budget Lab at Yale has tracked this by income group since the current round of tariffs began, and the pattern is consistent: households in the bottom income decile carry a tariff burden of roughly 1.3 percent of their after-tax income under the current policy path. Households in the top decile carry about 0.4 percent. The bottom decile’s burden runs close to three times the top decile’s, as a share of income.

Chart note: this figure is sourced from an academic research center, not a government statistical agency — see the chart's own source block for the full sourcing note.

The dollar amounts tell the same story from a different angle. The bottom decile pays about $517 a year under the current path; the top decile pays about $2,175. The dollar figure is larger for the wealthier household. The bite out of the tank is not.

Fuel tax runs on the same logic. Every gallon carries the same tax, whether it’s going into a work truck a household depends on to get to a job or a second car nobody needs to drive. Research on gasoline spending finds it eats up roughly 10 percent of income in the lowest income group, against roughly 2 percent in the highest. One caveat worth naming plainly: measure the same tax against what a household spends instead of what it earns, and the gap narrows. Both are legitimate ways to look at the same number. This piece uses the income-based measure because it matches how the other figures here are framed — but the fact that two honest measurements tell different-sized stories about the same tax is itself worth sitting with.

Occupational licensing works differently but lands in the same place. Getting permission to work in dozens of lower-income trades — contractors, cosmetologists, home health aides, and similar roles — requires, on average, nearly a year of required training or experience, one exam, and more than $260 in fees. A worker who already has less time and less money to spare is asked to spend more of both just to be allowed to earn.

Court fines and fees complete the picture. Close to one in five working-age adults reported a household fine or fee in 2024, and the burden falls hardest on families connected to court or incarceration costs specifically — not routine parking tickets, which are spread fairly evenly across income levels. About six percent of adults carried an unpaid fine or fee balance as of the end of 2023, and one in four of those owed $1,000 or more.

Tolls belong in this same family, though the honest answer here is smaller than the others: the pattern — a flat per-trip charge, same for every income level — is well established in the research on road funding. A clean, comparable dollar figure isn’t. Rather than force a number that doesn’t exist, it’s worth naming the shape of the problem and leaving it there.

None of these taxes were designed by the same office, passed in the same year, or aimed at the same target. A city sets tolls. A state sets fuel tax and licensing fees. A court sets fines. A trade policy sets tariffs. Nobody sat down and decided to build a matching set of regressive charges. But because each one is flat instead of scaled to income, every single one produces the same result: a smaller bite for the large tank, a larger bite for the small one.

That’s the piece that doesn’t show up on any one agency’s ledger. Tariffs get counted as trade policy. Tolls get counted as transportation revenue. Licensing fees get counted as consumer protection. Fines get counted as court administration. No one adds them into a single number and asks what a flat-cost system, taken as a whole, actually does to a paycheck with less room in it. The toll gets measured five different ways in five different offices — and never once as the same tank losing air five different times.

The diver with the small tank isn’t imagining the panic. The gauge on someone else’s larger tank just never had to move.

Every aircraft that uses the runway pays a fee scaled to its weight. Almost every aircraft.

Here’s a piece of everyday infrastructure almost nobody thinks twice about. Airports charge landing fees, scaled to the weight of the aircraft and how heavily it uses the runway. That fee helps fund the maintenance, the fire and rescue crews, the whole operation that keeps the tarmac safe for every other plane using it too. Nobody calls this radical. It’s just understood: if your aircraft puts more strain on the system, you pay more to keep that system running.

That’s the deal almost everywhere in this country, at almost every scale. A shipping company pays port fees scaled to the size of the vessel and the cargo it’s moving. A trucking company pays road-use fees tied to the weight its rigs put on the pavement. A utility customer pays a hookup fee sized to how much capacity they’re drawing from the grid. The bigger the load, the bigger the fee. Nobody’s exempt from that logic, until you get big enough.

In October 2008, Congress passed the Emergency Economic Stabilization Act, authorizing up to $700 billion, later capped at $475 billion, to stand behind institutions whose failure threatened to take the rest of the financial system down with them. That’s an open-ended commitment of the country’s full financial backing, made available on short notice, to cover a scale of risk no private fund had ever been built to absorb.

Here’s something worth thinking about. An airport funds its fire and rescue crews through the landing fees collected from ordinary, everyday traffic. When a plane needs that crew for an emergency, the response is already funded by fees the whole runway’s users have been paying in all along, not assembled on the spot. In 2008, the emergency response that showed up for the country’s largest financial institutions wasn’t funded that way. It came from the country’s general taxpayers, the equivalent of every small regional flight suddenly covering an emergency landing by the heaviest jet in the fleet, one that had never paid a landing fee sized to its own weight in the first place.

The heaviest aircraft in the fleet never paid a fee sized to its own weight

The final numbers, once every program closed out fifteen years later, actually came in far better than the 2008 headlines suggested. According to the Government Accountability Office’s final audit, total disbursed was $443.5 billion, and the lifetime net cost after all repayments, sales, dividends, and interest was $31.1 billion, a fraction of the original number. The bank stabilization programs alone were profitable, returning roughly $24 billion more than they cost. The net expense came almost entirely from housing assistance for homeowners, money that was never meant to be repaid in the first place.

That’s genuinely good news about how 2008 turned out. It doesn’t change what happened at the moment the emergency response was needed. Nobody waives a landing fee retroactively because the plane happened to land safely after all. The point was never what the final repair bill came to. It’s that the heaviest aircraft in the entire system had never been paying a fee sized to the strain it put on the runway to begin with.

A companion piece on this site already laid out how flat, unscaled fees fall harder on smaller incomes simply because they’re flat. That’s one direction this problem runs. Here’s the other direction it runs, at the opposite end of the scale entirely: the single heaviest user of the entire financial system has never paid a landing fee sized to the strain its weight puts on the runway.

The Market Stability Assessment is that missing fee, in its simplest form. A constant 0.025 percent applied to qualifying market participation, funding the system’s upkeep the way a landing fee funds a runway, proportional to use, collected continuously, not assembled after a crisis. Investment, wages, and productive business capital aren’t touched by it at all. It’s aimed at the kind of concentrated financial activity large enough to matter systemically if it ever put real strain on the system.

That word constant matters more than it sounds like it should. Compare it to how Social Security's wage base works. That's a dollar ceiling, not a percentage, and it has to be raised by hand almost every year just to keep up.

Every one of those wage base increases required its own adjustment. A flat percentage doesn’t have that problem. It scales automatically with whatever it’s applied to, without anyone needing to update a number in a spreadsheet every January.

None of this would make the United States the first country to charge something like a landing fee for outsized use of its financial system. Hong Kong has charged a stamp duty on stock trades since the 1990s, funding roughly six percent of its entire government revenue, with no financial collapse to show for it. The UK’s Stamp Duty Reserve Tax has run at 0.5 percent since 1986. South Korea, India, Taiwan, and Switzerland all charge their own versions, in some cases for decades, in Switzerland’s case since the era of the First World War. France and Italy added theirs after 2012.

The one well-known failure is worth naming directly rather than avoiding, because the reasons it failed are exactly the reasons this design is built differently. Sweden tried this in 1984 at 0.5 percent, then doubled it to a full percent two years later, applied only to trades made through Swedish brokerages. Investors simply routed their trading through London instead. By 1990, roughly half of all Swedish equity trading had left the country, and the whole thing was repealed by the end of 1991. The rate was too high, and the loophole was obvious. Every version still running today, including the one proposed here, runs at a small fraction of Sweden’s rate and closes the exact loophole that sank it, assessing the security based on where the underlying company is domiciled rather than where the trade happens to clear.

There’s one more piece worth naming. Assessment revenue like this doesn’t disappear into general spending. It flows into the Stability Fund, the same fund built to respond if an institution or a sector needs support in the future. That means the next emergency response, if one is ever needed again, starts partially funded in advance by the aircraft whose weight creates the risk, instead of being assembled entirely after the fact, the way it was in 2008.

Every other heavy user of shared infrastructure in this country pays a fee sized to the strain it puts on the system. This is the one exception. Closing it isn’t a punishment aimed at success. It’s applying the same rule every other heavy user already follows, to the one place it was never applied at all.

The full participation-assessment framework, including the exact revenue modeling and how it interacts with every other funding stream, is written out in Title IX of the Alignment Economy Act and Appendix F of the Technical Appendix. Both are available now as full downloadable copies: the Legislative Framework and the Technical Appendix, $1.99 each. This piece is the plain-language version. The primary documents are the real thing.

Want the next one delivered instead of stumbled on? Subscribe at christinemarletti.substack.com

Rent, groceries, gas, insurance, childcare — every basket this project uses to measure the cost of living carries a dollar sign. Waiting doesn’t. The two hours spent in a chair for your name to get called never shows up anywhere.

That time isn’t free. It has to come from somewhere — a shift you didn’t work, a kid you couldn’t pick up, a nap you didn’t get. It doesn’t show up in the official poverty line, in household budget calculators, or in this project’s own cost basket. Time spent waiting is a real cost of being poor, and no gauge currently measures it.

Here’s what we do know, from a study built to measure exactly this. Economists Stephen Holt and Katie Vinopal used the government’s own time-diary data to compare how much of an average day gets spent waiting for basic services — a line at the pharmacy, a hold with a call center, a seat in a waiting room. Relative to high-income people, low-income people are more likely to wait at all, and when they do wait, the wait runs about 12 minutes longer. Add it up over a year and low-income people spend roughly six more hours waiting than high-income people do — nearly double the total. The study controlled for the obvious explanations — family obligations, education, how far people commute, how many hours they work. The gap remained even after accounting for all of it.

Some of that waiting happens at the exact places built to help. The Social Security Administration’s own numbers show the same gap in miniature. In May 2026, people doing business online waited zero minutes. People who called and reached an agent waited about five. People who had to show up in person waited twenty minutes on average — often because they lack reliable internet, the online system won’t take their case, or they don’t trust it, and often during work hours they weren’t free to miss.

Waiting itself is only half of it. Federal law sets a firm deadline for how fast a SNAP application has to move: 30 days for a regular case, seven for an emergency. But the actual state-by-state record shows wide gaps in whether that deadline gets met — in some states the large majority of applications clear on time, in others a bare minority do, for the same federal program under the same federal law. Whether that legal promise gets kept depends heavily on which state a person happens to live in.

Missing that window has consequences that compound. When states tighten verification and recertification requirements — more paperwork, more frequent check-ins, more proof required more often — that burden falls on people trying to stay enrolled in a program they already qualify for, not just people applying for the first time. Recent enrollment data shows SNAP participation fell by roughly 4 million people between mid-2025 and early 2026, even as unemployment stayed essentially flat. When the number of people needing help doesn’t change but the number receiving it drops that fast, the paperwork is doing some of the cutting on its own.

Time scarcity doesn’t just cost hours. It costs money directly, in ways that don’t announce themselves as a “time” cost at all. Research on household grocery spending found that when the person doing the shopping had less time and less practice at it, the same cart of groceries cost about 5% more. Price-sensitivity is a skill built through repeated, unhurried trips. It has nothing to do with how careful the shopper is. It erodes fastest in whoever has the least time to practice it.

None of this is anyone’s fault, and none of it requires a villain. Call centers get understaffed when budgets get cut. Verification requirements tighten when someone upstream needs to hit an error-rate target. Add those small decisions up and the total comes to six extra hours a year for the people who can least afford to lose them.

A cost of living that leaves out the clock isn’t measuring the whole cost.

The game at the carnival looks like it has a solution. You have a mallet. You have a mole. You hit it. For a moment, the mole is down. Then another one pops up somewhere else. You hit that one. The first one comes back. The game doesn’t end because you’re winning. It ends because you run out of tokens.

That’s a reasonable description of federal housing, healthcare, childcare, energy, and transportation policy for the last forty years.

The problem isn’t that nobody is trying. Every intervention listed below represents real effort, real legislation, real votes. Some of them have helped real people. The problem was baked in from the start: when you fix one component of the basket without anchoring all of them to what wages can actually cover, the pressure doesn’t disappear. It moves.

Whack one mole. Watch the others.

The basket and the mallet

The essential basket — the minimum annual cost for one adult to remain economically functional — has six components. Housing. Food. Healthcare. Transportation. Utilities. Communication. Together they add up to approximately $36,000 per year at current costs. The federal minimum wage, frozen since 2009, produces about $15,000 per year before taxes. Even double that floor comes up short.

Each component has been targeted by federal policy at least once in the last four years. Here’s what happened each time.

What the pattern shows

Look at the grid and a pattern becomes visible. Every intervention is real. Every vote was cast. Every program reached someone. None of them are connected to each other. None of them are connected to wages. And in several cases, fixing one component created pressure in another — drug price caps produced higher launch prices, energy credits produced savings for homeowners while renters kept paying full rates, infrastructure investment produced future capacity while current households paid record auto loan costs.

This isn’t an argument that the legislation was wrong. The housing bill addresses a real supply problem. The IRA drug provisions save real money for real Medicare beneficiaries. The infrastructure bill builds real roads. Each intervention makes sense on its own terms.

The problem is systems design. In a connected system, everything affects everything else. When one component is adjusted without alignment, pressure moves rather than disappears. The book puts it this way: relief solves immediate pressure. Resolution fixes what’s underneath it. Most of these interventions deliver relief. None of them deliver resolution. The system feels better for a moment, then drifts again.

What resolution actually requires

The game at the carnival doesn’t have a winning strategy that involves a faster mallet. The game is designed so that speed doesn’t win. More tokens don’t win. The only way to stop playing is to change the game.

What the basket needs isn’t a better intervention on any individual component. It needs an anchor — something that connects all six components simultaneously to what the economy is actually producing per person per hour, so that when productivity grows, the basket costs and the wages tracking them move together rather than drifting apart.

That anchor has a name: the National Productivity Unit, the framework this book is built around. It ties wages, the cost of the basket, and what the economy actually produces into one connected number, automatically — so a housing bill doesn’t quietly push the pressure into someone’s drug costs instead.

Without that anchor, the pattern in the grid above will continue. A new housing bill will pass. A new drug will get price-capped. The broadband program will get refunded. And the moles will keep coming, because the game is still running and nobody has changed the machine.

Not because the people swinging the mallet aren’t trying. They are. But you can’t win this game with a faster swing. You have to change what the game is measuring.

This week, one number made headlines. Initial jobless claims fell to 187,000, the fewest since September 1969, before most of today’s workforce was born. Some outlets ran it as proof of a booming economy. Others ran it straight, a genuinely strong reading with a caveat or two buried in paragraph six.

Both readings used the same real number. Neither one is wrong. That’s the part worth sitting with.

Looking at just the headline rate is a little like watching a play through a cardboard tube pointed at the lead actor. He’s hitting every mark. He looks great. Pull the tube away, though, and you can see two supporting actors already walked off set, a stagehand is scrambling to hold up a backdrop that’s starting to tear, and the orchestra pit is quietly taking on water. None of that shows up if you only ever look through the tube.

The same week, the same federal government published five other numbers. None of them made the headline.

Start with who’s actually working. The share of Americans ages 25 to 54, prime working years, not retirees, not students, who are in the labor force fell to 83.3% in June. That’s a six-tenths drop in a single month, one of the sharpest this cycle, and it wasn’t concentrated at either end of that age range. This isn’t the Baby Boomer retirement story people reach for by habit. These are people in the middle of their working lives, and fewer of them are working or looking.

The government’s own broader unemployment measure sits at 7.9%. It exists specifically to count what the headline rate misses: part-time workers who want full-time hours, people who’ve stopped actively searching but still want a job. That’s nearly double the 4.2% everyone quotes.

That 4.2% is worth a second look on its own. A month earlier, it was 4.3%. The rate fell, and that sounds like good news, until you check why. 720,000 people left the labor force that month. Actual employment, people working, fell by 507,000 over the same stretch. The rate went down because fewer people were being counted, not because more people found work.

Long-term unemployment tells a similar story from a different angle. 1.9 million people have now been out of work for 27 weeks or more, more than one in four of everyone currently unemployed. That number is up 286,000 from a year ago. Under most state rules right now, once someone exhausts 26 weeks of benefits, the extension programs built for exactly this situation aren’t triggered. They need a state unemployment rate most states aren’t currently near. So the exhausted worker doesn’t get an extension. They just stop showing up in next week’s count.

None of this makes the 187,000 number false, and none of it means anyone reported dishonestly this week. The number is measuring exactly what it was built to measure: how many people applied for a specific benefit in a specific week. It was never built to measure whether someone gave up before applying, exhausted their benefits months ago, or left the workforce entirely rather than keep looking. This is a description of what happens when a country has one headline number and five supporting ones, and only the headline gets read out loud.

Financial markets, for what it’s worth, are barely registering any of this. The Federal Reserve’s own stress index, a composite of interest rates, credit spreads, and market volatility, has read below-average stress for five straight weeks. Wall Street’s instruments and a family’s instruments are measuring two different economies, and both readings are accurate.

A country that wants an honest answer to “how’s the job market doing” needs more than one number that gets read on the news and five that don’t. It needs all six read together, on purpose, every time, not assembled by hand, after the fact, by whoever happens to go looking.

Until that’s built, whether the economy “feels” strong this week depends on which of the six numbers you happened to see first.

Somehow that logic disappears the moment the tool is AI.

Right now, a genuine industry has formed around catching people for using it, and I think that’s a waste of everyone’s energy. Detection software gets sold to schools and employers with confident marketing — 98%, 99%, sometimes “99.5% accurate.” Independent testing tells a different story. Across real benchmarks, no detection tool clears 85% accuracy, and false-positive rates — flagging a real human’s real writing as machine-generated — run anywhere from 2% to 12% for native English speakers, and considerably worse for people who don’t fit the tool’s assumptions about what “normal” writing looks like. One 2026 study found detectors misfired on 61% of essays from Chinese students, against about 5% for American students, using the exact same test conditions. Non-native English speakers and neurodivergent writers get flagged at the highest rates of anyone — punished not for using AI, but for writing in a way a piece of software didn’t expect.

The response from serious institutions has been to back away. At least sixteen major universities — Yale, Vanderbilt, Johns Hopkins, Northwestern, UCLA among them — have disabled AI detection tools entirely. OpenAI shut down its own detector after admitting it didn’t work. When the people who built these systems and the people paying to use them are both walking away, that tells me something: the tool doesn’t do the job it’s sold to do.

Meanwhile, in workplaces, the pattern is even more telling. Recent survey data shows nearly half of employees hide their AI use from their employer — not because they’re cutting corners, but because disclosure carries a real cost. One controlled study had people evaluate the exact same piece of work, with the only difference being whether it was labeled as AI-assisted. The AI-labeled version got rated ten times lazier and was 24 percentage points less likely to get recommended for a high-visibility project — same work, same quality, worse judgment of the person who produced it. That’s bias against the tool, dressed up as a standard for the work.

I’ve spent thirty years in plant operations. I’ve watched a lot of tools come and go, and I’ve watched the same fight happen every time. Calculators went through this exact argument in the 1970s and 80s. Teachers said they’d rot basic math skills. Some states banned them from standardized tests. By the 1990s calculators were standard equipment, and nobody looks back at that panic and thinks the resisters were right. The tool changed what people were free to spend their thinking on.

A cabinet that closes flush was built correctly, regardless of which drill built it. The same goes for an argument that survives scrutiny. I care about whether the seams hold. Not what made them.

If a piece of writing is wrong, tell me what’s wrong with it — the claim that doesn’t hold, the source that doesn’t check out, the logic that doesn’t follow. A right conclusion is still right regardless of what helped get it there. Judging a conclusion by the process that produced it, instead of by whether it’s true, is how bad ideas survive and good ones get thrown out for the wrong reason.

Debate the argument with me.

Three trucks left for the same place. None of the drivers talked to each other first.

One took I-90 east out of Seattle, through Montana, past the Livingston exit for Yellowstone. One took I-70, straight through Topeka and Kansas City. One took I-40, south through Albuquerque and out across the New Mexico desert. Same freight in every trailer. Same sticky note on every dashboard: Destination, Port of Miami.

Three different roads, three different reasons for picking them. None of the three drivers knew the other two existed.

The first driver runs the northern route because it tells him what’s coming before it arrives. That’s the argument coming out of EPB Research this month: the big numbers everybody already leans on — GDP, the unemployment rate, payrolls — are also the numbers that define a recession, not the numbers that warn one is coming. By the time they move, the thing they’re measuring has already happened. The numbers aren’t wrong. They just take longer to show up than a warning needs to.

The second driver runs the middle route because he doesn’t fully trust the gauge on his own dashboard. Kevin Warsh, the sitting chair of the Federal Reserve, said as much about the government’s own preferred inflation number this year: the PCE index, he told lawmakers, “offers only a rough take” on what prices are actually doing, even with the volatile stuff stripped out. That’s the person running the building, saying his own main gauge needs a second read.

The third driver runs the southern route because he’s decided the whole dashboard is pointed at the wrong thing. In May, the United Nations released a report built around exactly that argument: GDP was designed as a narrow measure of economic output and became something it was never built to be — the one number every country reaches for first. Secretary-General António Guterres didn’t call it a mistake. He called it “a longstanding blind spot in measuring progress.”

Each one wants a different fix. The forecaster wants a narrower, faster indicator. The Fed Chair wants a different inflation formula. The Secretary-General wants a global dashboard beyond GDP. Three different fixes. But underneath the fixes, all three said some version of the same thing: the number on the dashboard and what’s actually happening in front of the windshield have quietly stopped being the same number.

There’s a fourth route that isn’t in the photo. It runs through this book, The Alignment Economy, and it’s been making the same case since before any of them said a word about it this year: that pay and productivity stopped being measured against each other decades ago, and every dashboard since has been able to read green without ever checking whether the two lines are still connected.

None of the four needed to compare notes to end up saying some version of the same thing. The freight was always headed to one place, whether anyone said so out loud or not.

Right now, four different people can pull into the same port and describe it four different ways, because nothing yet tells them, in the same words, what they actually found.

That’s the plain part of the work left to do — not inventing a new destination. Just putting a sign on the one that was already there, so the next driver doesn’t have to arrive by accident.

_{Detroit at its industrial peak — the same decades the city was also being mapped, block by block, for credit risk}

In the 1930s and 40s, Black Bottom and Paradise Valley were two of Detroit’s most vibrant neighborhoods. Black Bottom was residential, dense, and tightly knit. Paradise Valley was its commercial heart — jazz clubs, Black-owned restaurants, doctors’ offices, and theaters lined Hastings Street, the corridor that tied the two neighborhoods together. Thousands of families and hundreds of businesses called it home.

Hastings Street, Detroit, Oct. 7, 1949 — 1054–1058 Hastings. Courtesy of Detroit Public Library. (colorized)

In 1939, the federal government drew a map of Detroit. The Home Owners’ Loan Corporation graded every neighborhood from A to D, ostensibly to help banks judge mortgage risk. Black Bottom and Paradise Valley were graded D — “hazardous” — and the grading notes for D-rated neighborhoods routinely cited the race of the people who lived there as the reason. Banks used the map for decades afterward to decide who could get a home loan and who couldn’t.

A 2021 study published in the American Economic Journal found that the HOLC maps didn’t just describe existing conditions — they caused what came after. Comparing neighborhoods on either side of a map boundary, researchers found the graded side experienced measurably lower homeownership, lower home values, and more segregation for decades, changes that tracked the map itself rather than anything that predated it.

By the 1950s and 60s, Black Bottom and Paradise Valley were gone. Urban renewal projects and the construction of Interstate 375 demolished Hastings Street and, with it, the commercial core that had held both neighborhoods together. The freeway that replaced it is just over a mile long — one of the shortest interstates in the country.

The 606 Horseshoe Bar, 606 East Adams Avenue — in the heart of Paradise Valley. Courtesy of Detroit Public Library. (colorized)

That gap still shows up in the numbers today.

None of this is unique to Detroit — Michigan has the highest state-level Black unemployment rate in the country as of the first quarter of 2026 — but Detroit is where the ninety-year-old map and this year’s numbers line up most exactly.

At the same time, Detroit is, by the headline measure, doing better than it has in decades. The city’s population grew for two consecutive years as of the 2025 census count, the first sustained growth since 1950. Investment is real and visible: a $1 billion mixed-use development on Woodward Avenue, JPMorgan Chase and General Motors relocating offices downtown, tax values climbing. National outlets have run the comeback story more than once.

Researchers at Michigan State and Wayne State looked at where that comeback is actually happening. They found it’s concentrated in about seven of the city’s 139 square miles — the downtown and Midtown core. Citywide, roughly one in three Detroiters still lives in poverty, the highest rate since 2017. A population count going up and a poverty rate staying flat are both true at the same time, and neither one, on its own, tells you whether the other is happening. That’s the trouble with a single headline number: it can read as entirely good news and still miss most of what’s actually going on a few miles away.

The clearest place to see the gap between the map and the recovery is happening right now, in the same footprint Black Bottom and Paradise Valley used to occupy.

_{MDOT’s conceptual rendering for the I-375 boulevard conversion, in the same footprint the freeway once cleared.}

The Michigan Department of Transportation has a project underway to convert Interstate 375 back into a street-level boulevard — the same highway that erased Hastings Street seventy years ago. MDOT’s own project materials describe the original construction as clearing “red-lined” neighborhoods and say the department is committed to honoring that history as part of the rebuild.

The project has not been a smooth, top-down fix. MDOT paused it in August 2025 after residents pushed back on the design and costs kept rising. The department came back four months later with a smaller, cheaper plan — down from roughly $520 million to $120–140 million, with a shorter boulevard and construction split into phases so residents would have more chances to weigh in before the concrete gets poured. The next public meeting is scheduled for March 2026. Mary Sheffield, sworn in as Detroit’s first woman mayor in January 2026, said during her campaign that she wants “more conversations about repairing the harms done to generations of Black families” by the neighborhoods’ destruction. Some residents, organized as the ReThink I-375 Coalition, still don’t think they’ve been heard enough on what gets built in that footprint next.

None of that makes any single person in this story a villain. A federal agency drew a map in 1939. A state transportation department is currently revising a construction plan because the people it affects told it to. A newly elected mayor is naming a history that’s ninety years old. What connects all three isn’t intent — it’s that the numbers everyone actually tracks (population, tax base, investment dollars) were never built to show whether a recovery reaches the ground a map once drew a line around.

She has been carrying plates since before you were born. The diner opened in 1965 and she was there on the first day. Same walk from the kitchen window to the corner booth, same coffee refills, same order pad, same conversation with the regulars who come in on Tuesday mornings because Tuesday is when the pie is fresh. The job has not changed in any meaningful way in sixty years. The plates are a little lighter now. The coffee machine is different. Everything else is the same.

For the first thirty years she worked there, her paycheck grew. Not dramatically. Not all at once. But steadily — the way things used to grow when the economy was building something and the people building it got to keep some of what they built. By 1995 she had raised two kids. She owned a house, eventually. Not a big one, but hers. She had enough.

The job didn’t change. The paycheck did.

And then, somewhere in those middle years, it stopped keeping pace. Not all at once — gradually, the way a bridge sags slowly before it shows the strain. The diner kept filling up. The coffee kept getting ordered. She kept walking from the kitchen to the corner booth. But the distance between what she earned and what a life cost started opening up — slowly, then faster, then too wide to close on a waitress’s income without help from somewhere else.

Nothing changed about the job. The connection between what she produced and what she received came loose. Like a drive shaft that stops transmitting — the engine keeps running, the work keeps happening, but the motion stops reaching the wheels.

In 1994 I walked the floor of a manufacturing plant in Long Beach and stopped to talk to a young man named Ronnie. He was maybe twenty-three, twenty-four. He built airplanes for a living. I asked how things were going. He said life was great. He had a great girlfriend, a job building airplanes, and twenty-two pairs of jeans.

Twenty-two pairs of jeans was Ronnie’s unit of measure for a life that was working. He showed up, he did skilled work, and the economy gave him something back that he could see and count and hang in his closet. The connection between what he produced and what he received was intact. He could feel it. He measured it in denim.

That’s what alignment looks like when it works. Not wealth. Not excess. A paycheck that tracks the value of the work well enough that a twenty-three-year-old on a factory floor can measure his prosperity in blue jeans and tell his boss about it with a straight face.

The economy kept producing more. The paycheck stopped tracking it. Not because anyone took something from Ronnie. Because the part that kept the two in step came loose, and nothing was built to catch it.

The job didn’t get worse. The connection came loose.

This sounds like redistribution. Redistribution is a verdict. It decides who deserves the gains after they’ve already landed somewhere. What a reconnected system does is different — it sets a metric so that pay and output move together as the output is being made. No verdict on anyone. A metric, not a judgment.

You’d pay people who won’t work. The metric doesn’t read anyone’s character. It reads what the economy produced. The waitress gets more when the economy produces more. She gets less when it produces less. Her effort is still the thing that earns her place at the table.

The waitress gets a raise because a factory owner bought a robot? She already did. For thirty years she got raises while no robot touched her diner. The efficiency gain at the plant two towns over lifted her pay, because the economy was producing more and the connection carried some of that back to her. The robot is just the newest version of every gain that ever came through.

The money will just leave. Capital goes where it’s treated best. The system runs on what the economy produces here. And here is the thing about a robot that makes goods nobody can afford: it’s a very efficient way to make things that don’t get bought. The factory owner needs customers. The customers are the workers. The reconnected system isn’t the owner’s cost. It’s his demand.

That was one argument wearing four hats. All four rest on the same buried premise — that your pay comes from your station. The waitress in 1965 disproves it. Ronnie in 1994 disproves it. The productivity-pay gap since 1979 disproves it at scale.

She is still there. The Tuesday pie is still fresh. The regulars still come in. The job is exactly the same as it was in 1965. What changed is whether the economy's output finds its way back to her in a paycheck that tracks what the whole system is producing. For thirty years it did. Then the connection came loose, and for the next thirty years it didn't — not reliably, not automatically, not as a matter of design. The gap opened. She absorbed it.

Picture her pay stub, years from now, after the part gets built. The dollar amount is still there — nothing about that changes. But underneath it sits a second number: what that hour of work costs against what the economy is actually producing right now, not what it produced last year, not what a committee argues it should produce. She doesn’t need the formula to read it. She glances down the way she’d check a thermometer, and she knows, before she even counts her tips, whether this week kept pace or fell behind.

The book calls that number NPUX — shorthand for expressing anything, a wage, a coffee refill, a slice of pie, as a share of the National Productivity Unit, the NPU. The NPU itself isn’t money. It’s a measuring stick, the same way a thermostat setting isn’t heat — just the number the system holds steady around, recalculated as the economy actually performs, not argued over in Washington once every few years. The diner’s specials board still shows the price of the pie in dollars. It’s just no longer the only number on it, and it’s no longer lying by omission about whether that price is keeping pace with what she’s paid to serve it.

None of this requires the diner to change how it prices pie, or requires her to negotiate her wage any differently. What changes is that everyone at the counter — the waitress, the owner, the regular paying for his coffee — is finally looking at the same instrument, one that moves when the economy moves, instead of one that only moves when someone remembers to update it.

That’s the part that was missing the whole time. Not a redistribution. Not a program. A shared number, sitting right next to the price of the pie, that tells you in real time whether the connection between the work and the wage is still holding.

It was never a dozen arguments. It was one loose connection — and a loose connection can be fixed, once you stop calling it a character flaw and start calling it what it is: the part of the design we never finished.