But buried in the text of a proposed constitutional amendment is something far more consequential: a tax structure that could make it harder for ordinary employees to take equity in startups, discourage founders from building in California, and ultimately do little to solve the healthcare funding problems it was supposedly designed to fix.

The measure is being sold as a one-time tax on extreme wealth. The reality is much messier.

California’s Billionaire Tax Isn’t Really About Billionaires

The 2026 Billionaire Tax Act, a California ballot initiative promoted by the Service Employees International Union–United Healthcare Workers (SEIU–UHW), seeks to tax billionaire net worth rather than income. As written, the measure would amend California’s constitution by removing the state’s existing cap on taxes on intangible personal property. Supporters argue that the tax is necessary to address a large shortfall in funding for healthcare programs.

The act claims that this will be a “one time” tax. But as my colleague Veronique de Rugy has pointed out, a billionaire tax like this won’t address the core issues at the heart of California’s healthcare funding problems: the core issue of state hospitals providing services via Medi-Cal, California’s Medicaid Program, to middle-class Californians and illegal aliens — both groups that shouldn’t qualify for assistance under the program. The factors driving the shortfall are structural, and a tax on wealth cannot address them.

If California’s healthcare spending commitments are unsustainable, the solution must involve reforming those commitments, not simply searching for a new source of tax revenue. The California state government would be better served to more carefully monitor Medi-Cal program participants to ensure that only the poorest Californians – the group originally intended to receive benefits – exclusively qualify. My colleague Jack Salmon points out that in 2015 there were roughly 12 million people on Medi-Cal. Today that figure is about 15 million people or roughly 4-in-10 Californians (a generous figure that includes Medicare claimants). Alternatively, the California state government needs to face some credible commitment that the funding for the Medi-Cali program provided via the California taxpayer can’t exceed the total revenue collected. Jack points out that “California doesn’t actually need new revenue to address the gap that would remain after accepting the OBBBA’s eligibility defaults.” And whether either of those options work, the Federal government could do its part to withhold funds typically dispersed as a match to state Medicaid funds.

One Provision Deserves Particular Attention

The measure states: “For any interests that confer voting or other direct control rights, the percentage of the business entity owned by the taxpayer shall be presumed to be not less than the taxpayer’s percentage of the overall voting or other direct control rights.”

It further provides:

“Notwithstanding any other provisions of the Constitution, the Act provides for taxation on all forms of personal and wealth, whether tangible or intangible, and allows for the classification of personal property and wealth for differential taxation or for exemption, for the purpose of imposing the on-time tax on the wealth of California billionaires. Personal property and wealth that is so taxed includes, but is not limited to, shares of capital stock, bonds or other evidences of indebtedness, and any legal or equitable interest therein.”

Provisions like this are significant for targeting wealth in forms like capital stock and intangible assets. Targeting intangible assets like equity endanger innovation as equity is typically included in compensation packages. Equity compensation awards individuals for the risk they take by giving them a stake in the company they work for. It aligns incentives towards shared goals of company growth regardless of the ultimate outcome -- whether that’s going public and seeking greater capital investment or seeking an acquirer. Taxes like these distort these incentives and inhibit productive potential as employers think twice about equity awards if employees demand higher salaries in anticipation of wealth taxes like this. This slows innovation. This slows economic growth.

Why Compensation Matters

Startup founders are often compensated through equity in the companies they build. This usually takes the form of Class A or Class B shares, which carry voting privileges not afforded to holders of other classes of equity. These are called super-voting shares.

Under California’s proposed billionaire tax, equity held by founders and their employees would be taxed because it would fall within the measure’s definition of “capital stock” and other intangible assets. Recent analysis by the Tax Foundation illustrates the potential burden for six of the top tech founders. While some have expressed no issue with the measure, it is clear they don’t speak for everyone.

One such dissenter is Google co-founder Sergey Brin, whose family sought refuge from the Soviet Union. In an interview with The New York Times, he offered perhaps the strongest public language against the proposal: “I fled socialism with my family in 1979 and know the devastating, oppressive society it created in the Soviet Union. I don’t want California to end up in the same place.”

Brin’s rebuke is born out of concern for a place of great affection and deep understanding of what it takes for young people to succeed and innovation to advance. He doesn’t point to the hardship of the rich under Socialism, but all people. A policy like this has far reaching consequences beyond billionaires as their middle class employees shoulder burdens they can’t bear.

Equity Isn’t Cash

It should also be said that not every founder is a billionaire with spare cash to plug the crater of fiscal irresponsibility created by the state of California.

Taxing equity assumes that shares are liquid enough to generate the cash necessary to pay the tax. In many cases they are not. Taxing equity would be catastrophic, especially for the middle class involved in new ventures.

The Billionaire Tax Hurts the Middle Class

There are a few reasons for this. One is that it’s bad for talent recruitment.

Founders are not the only ones who are awarded equity as a part of their compensation packages: Non-founder employees pour their lives into the companies they work for, devote their time and talents to building something new that takes long nights, weekends and bends intellectual capacity well beyond the average worker (and as an average worker, there’s nothing wrong with being average!). Their passion powers their work.

One such example of innovative companies making average and often unsuspecting employees millionaires is Space X, which went public late last week. The tough, risky, and innovative work of Founders who have the need and capacity to employ large groups of people with necessary skills enables people like Juan Hernandez, a welder at Space X, to become a millionaire when the company went public. While Space X isn’t headquartered in California, Elon Musk’s other companies once called The Golden State home. It’s painful to imagine all the other people like Juan who are in California and are hampered in their ability to build generational wealth because irresponsible bureaucrats refuse to reign in reckless healthcare spending. Innovation and the strength of American capital markets work together to make the American dream a reality for everyday people regardless of where one starts out. It’s important that the American people aren’t penalized for simply participating in our markets.

Taxing equity in this manner gives innovators pause when deciding to start a new venture and adds friction to a potential employee’s decision to join a new venture in a way that shouldn’t be a second thought. Withholding the capacity for a share in the project they’re responsible for building drives high value talent away from ambitious projects and stalls economic growth.

The other reason this supposed billionaire tax will likely fall on middle-class workers is that ultra-high-net-worth and high-net-worth individuals possess the capacity to not only move their money but move themselves and their productive efforts to other places entirely.

We have repeated evidence to suggest that efforts like these in other blue states drive out the state’s tax base. My colleague Jack Salmon has pointed out the track record of such policies and the impact here.

Moving to different states or countries affords founders and potential employees more amenable environments that allow businesses to not only select their compensation arrangements to attract top talent, but to avoid uncertain and volatile business environments that endanger aspects of their core business (i.e. any kind of regulation on energy, environmental usage, etc.).

We’ve Seen This Before

This isn’t the first time our nation has flirted with a tax proposal targeting high earners.

The Affordable Care Act of 2010 included a provision for a 3.8% net investment income tax (called the ACA surcharge tax). This tax broadly applied to filers with incomes of $200,000 for single filers, or $250,000 for couples filing jointly. This tax included unearned income including investment income such as taxable interest income, dividends, realized capital gains, nonqualified annuities, rents, royalties, passive income from business activities, and undistributed net investment income from a trust or estate.

Worse yet, like the ACA Surcharge Tax, this proposed tax isn’t inflation adjusted either. In the case of the ACA Surcharge Tax, that means that eventually, middle-class people — people intended to have been excluded from the burden — will end up paying it. Jack Salmon discusses this in depth here.

The Billionaire Tax Can’t Solve the Spending Problem

The truth is this: When government projects promise services funded by other people’s money, the spigot of taxpayer funding can’t stop. And what starts as a promise of a running faucet can quickly become a firehose when programs lack accountability and core issues are never addressed.

That is exactly the challenge California faces today. Continuing to tax average earners will not solve the state’s healthcare funding shortfall, and dreaming up new taxes targeting billionaires — a category too broad to avoid — is likely to hamper economic growth. California’s proposed billionaire tax risks discouraging investment, entrepreneurship, and innovation: the very activities that generate economic growth and prosperity. This is true for individuals, the state of California, and as the popularity of taxes like this grow, the nation more broadly.

Innovation and the ability to take risks and receive one’s due reward for such risk taking is integral to the American spirit and to an American growth strategy. So-called billionaire taxes endanger the progress that grows the economy for everyone and enables irresponsible government spending.

A new working paper by David Neumark and Antonio Rodriguez-Lopez, “Modern Difference-in-Differences, Same Old Answer,” lands a serious blow against recent high-profile research claiming that minimum wages have little to no effect on jobs. Their argument isn’t that event-study designs are bad. It’s that the null results attributed to these methods are artifacts of specific, defensible-sounding, but ultimately flawed researcher choices that, once corrected, yield the same old answer: minimum wage increases reduce employment.

The Methodological Sleight of Hand

The papers in the crosshairs are Cengiz, Dube, Lindner, and Zipperer (2019), known in the literature as CDLZ, and a recent Handbook of Labor Economics chapter by Dube and Lindner (2024). Both of these are influential studies. CDLZ’s stacked event-study design has become something of a gold standard in the minimum wage literature, cited over 3,000 times, with a majority of citations specifically referring to the stacked design. Dube and Lindner’s handbook chapter has helped cement the narrative that modern diff-in-diff methods “correctly show a null effect.”

Neumark and Rodriguez-Lopez, working with the same data, the same period coverage, and the same event-study architecture, identify several choices embedded in the null-results papers that are doing a lot of work.

The most striking is the dependent variable. CDLZ measure employment as a share of a varying population, that is, the current population in each period, which itself changes in response to minimum wage policy. This matters enormously. There is a well-documented negative relationship between minimum wages and population: higher minimum wages tend to reduce population in affected areas, as workers and firms relocate. When both employment and population fall after a minimum wage hike, the ratio of the two can actually increase, making it look as though jobs were created when in fact jobs were destroyed. CDLZ never justify this choice, and it’s inconsistent with their stated goal of estimating effects on low-wage jobs.

When Neumark and Rodriguez-Lopez switch to either a constant population denominator or simply log employment, both more defensible specifications, the long-run employment elasticity shifts from a statistically negligible +/- 0.01 to a significant -0.09 (weighted) or -0.20 (unweighted).

The Range Problem

The second major issue is sample truncation. CDLZ’s stacked design focuses only on jobs in the four-dollar range below and five-dollar range above the new minimum wage. The implicit assumption is that minimum wage effects are zero outside this band and that firms don’t respond to a wage floor by cutting jobs across the broader wage distribution. Neumark and Rodriguez-Lopez show this assumption is wrong. When firms exit the market due to minimum wages, jobs are destroyed along the entire wage distribution, not just in the directly affected wage bins. Including employment effects outside the restricted range yields significant additional job losses.

Restaurants: Even Harder to Dismiss

The restaurant sector has long been the canonical case study in minimum wage research, precisely because it employs many low-wage workers and faces thin margins. Even the null-results literature concedes something here: Dube and Lindner’s own estimates show a weighted own-wage elasticity of -0.03 for restaurants.

Neumark and Rodriguez-Lopez find something more striking: this estimate is so fragile it collapses under the weight of a single methodological choice. Switch from weighted to unweighted estimation and the restaurant own-wage elasticity moves from -0.03 to a significant -0.38. The difference turns out to be driven by California and New York, two enormous, high-productivity, high-wage states that dominate weighted regressions. When those two states are excluded from the Dube-Lindner analysis, both weighted and unweighted estimates point to job losses. What’s true in California isn’t necessarily true in lower-wage labor markets, and those markets are where minimum wage increases hit hardest.

In the restaurants sample excluding California and New York, the employment-to-population ratio declines steadily and significantly in the years following a minimum wage event, reaching -2 to -3 percent by year four or five. In the full weighted sample, that signal is swamped by the gravitational pull of two states whose labor markets are sufficiently tight that minimum wage hikes may genuinely be less disruptive.

What This Is Really About

This paper is a case study in what we might call methodology laundering, the use of ostensibly rigorous modern techniques to lend credibility to findings that ultimately depend on contestable underlying choices. The stacked event-study design is not the problem; it’s a genuine methodological improvement in many respects. The problem is that null results have been presented as if they follow inevitably from the design, when in fact they follow from the design plus a moving denominator, plus a narrow wage window, plus a weighting scheme that amplifies the most unusual labor markets in the country.

None of these choices are obviously wrong on their face. Some even have superficially plausible justifications. But taken together, they systematically push estimated effects toward zero in ways that would not be expected under genuinely neutral research design.

Getting the employment effects of minimum wage right isn’t an academic exercise. States and cities are setting wages well above historical norms, often relying on research that, as this paper shows, is more fragile than its prominence suggests.

The same old answer, it turns out, was right.

The United States’ gross federal debt of 39 trillion dollars has become so incomprehensibly large that many citizens and politicians alike now simply choose to ignore it. To put 39 trillion dollars into perspective, if every citizen, baby and grandma alike, paid their share of the debt, we would each owe roughly $114,000. All the while, with this looming debt rain cloud overhead, the current administration’s spending cuts target relatively small discretionary programs while allowing the largest spending categories to balloon in size.

In FY2025 the federal government spent 7 trillion and brought in only 5.8 trillion in revenue. That 1.2 trillion-dollar gap is larger than the GDP of all but 19 countries. But instead of trying to rework the main drivers of debt such as entitlement programs like Social Security, Medicare, and Medicaid, our current government sticks to either targeting smaller, politically motivated cuts like the ones in the recent Rescissions Act of 2025 or performs fiscal judo in order to move spending from one area to another in disguise as budget cuts.

Compounding the problem, rising debt also leads to rising interest costs, forcing taxpayers to devote more resources simply to servicing past borrowing. Look below at how our government breaks down its own spending.

Source: cbo.gov

Social Security, Medicare, and Medicaid together account for roughly half of all federal spending, and their costs are projected to grow as the population ages and heath care cost increase, making them the primary drivers of long-term budget deficits.

The Rescissions Act of 2025 was put into place to cancel 9.4 billion dollars in funds that were provided to various government agencies for tasks that the Trump administration deemed unnecessary. The cuts primarily targeted foreign assistance and international programs, including Contributions to International Organizations, Contributions for International Peacekeeping Activities, Global Health Programs, Migration and Refugee Assistance, and the Complex Crises Fund. Notably, the rescissions avoided major entitlement programs, which account for the bulk of federal spending.

While 9.4 billion is no small number, it represents a very small portion of federal spending, and an even smaller amount of the federal debt. At current debt levels, these rescissions amount to shaving less than three-hundredths of one percent from the total federal debt burden.

Keep in mind that Social Security alone spent about $1.575 trillion in FY2025. At that rate, the entire $9.4 billion rescissions package equals about two days of Social Security spending. This highlights a larger problem in our current political system, which is a willingness to spend more, and an apprehension to any meaningful spending cuts.

This figure shows that even when the current administration ostensibly makes “spending cuts”, they just use the cuts to move money from one area to another Whether or not the temporary provisions of the OBBBA expire or are extended, it will add trillions of dollars to the debt. The act pulled funding from programs such as SNAP and Medicaid while dramatically increasing spending on immigration enforcement, border security, and other favored priorities.

Restructuring SNAP and Medicaid through spending cuts could be good policy, but not when you also go and spend 500 billion dollars largely on immigration and border enforcement, defense and farm subsidies. And not only did the OBBBA increase spending, but the bill also cut back taxes.

The result is that even after these cuts, the Congressional Budget Office estimates the omnibus bill will increase federal debt by trillions of dollars over the next decade. When spending cuts lead to an increase in the federal deficit and overall debt, they simply aren’t what they claim to be.

This isn’t to say the government shouldn’t look to cut costs anywhere possible. In fact, I’m generally for more rescissions. But this politically charged shaving of costs doesn’t get us meaningfully closer to working off our debt. Because instead of looking into how we can rework our entitlement programs, our government is more interested in pulling funding from the other parties’ programs. I believe this will only give motivation for the other side to do the same, creating an artificial cost-cutting war that ignores the real drivers of debt.

This may leave some of you wondering: why not just tax more? Well, the numbers show spending policy, as compared to tax policy, has a drastically greater influence on our debt problem. In a recent article by Jack Salmon, he found that “98% of the long-term structural deficit can be attributed to spending policy decisions, while just 2% is attributed to tax policy.” And of that 98%, the largest driver of the debt is Medicare. Mandatory spending programs are the main driver of our nation’s long-term structural deficit. Therefore, if our goal is to work towards lowering the deficit and eventually the debt, mandatory spending policy is the place to target.

Ultimately, while the Rescissions Act may save billions, America’s debt problem is measured in trillions. Instead of targeting politically motivated easy wins when it comes to spending, the real solution will require confronting the structural drivers of spending and having an honest conversation about the tradeoffs necessary to restore fiscal sustainability. Until then, lawmakers will continue celebrating small victories while the debt marches steadily upward, passing the costs of today’s political convenience onto future generations.

At the heart of the argument is the idea that the growth-centered economic model has run its course for poverty alleviation, and that growth itself should be discouraged for rich countries.

Many have noted how insane this idea is. Some have also rightfully labeled it an exercise in envirodictatorship. However, the offering of this idea has elevated the discussion about the importance of growth. In the course of reading about how people responded to Piketty, I found a new research paper that challenges the idea that growth alone isn’t enough, and that development aid or a global government to address global poverty is as necessary as growth.

Growth Is Enough, and Only Growth is Enough

Lant Pritchett of the London School of Economics and Addison Lewis of Brigham Young University have released a paper with a deceptively simple title: “Economic Growth Is Enough and Only Economic Growth Is Enough.“

Their target is a claim that has become fashionable in development economics: that growth is not sufficient to improve human wellbeing, that targeted programs and redistribution are “equally important,” and that poor countries should worry as much about the distribution of income as its growth. To this effect, Pritchett and Lewis cite a bunch of examples. The executive director of J-PAL, one of the most influential development research organizations in the world, put it baldly in a 2021 op-ed: “growth is not enough.” Yale’s Rohini Pande, director of the Economic Growth Center, wrote that growth “will not be sufficient to eradicate extreme poverty.”

This phenomenon, taken to its extreme, is what Piketty and his friends express when they argue against growth.

In that paper, Pritchett and Lewis set the record straight. They aim to prove something stronger about growth than a mere correlation: namely, that every general, plausible, cross-national measure of the basics of human material wellbeing has a strong, non-linear, statistically robust relationship with GDP per capita.

“Basics” here means what you’d expect: things like child mortality, nutrition, access to clean water and sanitation, shelter, primary and secondary schooling, and life expectancy. The floor of what a decent human life requires.

They construct multiple measures of the basics using entirely different methods, then compare them to GDP per capita. They also run three deliberate “data undermining” exercises to search specifically for indicators and weights that minimize their relationship to GDP per capita, to find the weakest possible result.

What they find is that regardless of how you measure the basics of human wellbeing (the Legatum Prosperity Index, the Social Progress Imperative’s Basic Human Needs index, the Oxford Multidimensional Poverty Index, or headcount poverty rates), the relationship with GDP per capita has four features that hold across every specification:

• The relationship is strong. GDP per capita alone predicts basic human wellbeing with a correlation of around 0.90. The statistical significance is, in the authors’ own word, ‘astronomical.’

• The relationship is non-linear. Growth matters most where countries are poorest. For the world’s bottom billion, an extra dollar of GDP per capita does two to eight times as much for basic human well-being as the same dollar does in a middle-income country. The poorer you are, the more growth transforms your life, which is precisely why capping it is most cruel to those Piketty claims to be helping

• Third, GDP per capita is empirically sufficient. There are no countries with high GDP per capita that have low scores on basic human wellbeing.

• Fourth, GDP per capita is empirically necessary. There are no countries with low GDP per capita that have achieved high scores on basic human wellbeing. You cannot get there without it. No combination of targeted programs, clever NGOs, redistribution schemes, or foreign aid has produced a poor country with rich-country health, nutrition, and education outcomes.

The lone outlier is Equatorial Guinea, a country that achieved high GDP per capita through oil extraction under a kleptocratic government that kept the gains confined to a narrow elite. The point is made: if growth is captured entirely by a tiny ruling class, it won’t flow to basics. But this is an argument for governance and institutions, not against growth.

Pritchett and Lewis do not just show that their preferred measures of basics correlate with GDP per capita. They run 100,000 simulations in which they randomly draw one indicator from each of seven basic domains of human wellbeing (health, education, nutrition, water and sanitation, housing, poverty, and natural environment) and combine them with randomly chosen weights. They then look for the specification that produces the weakest possible relationship with GDP per capita.

Even their worst-case result, which they achieve by rigging the index so that air pollution counts 40 times more than adult mortality and 29 times more than extreme poverty – a weighting scheme no serious researcher has ever proposed – nevertheless still shows countries getting meaningfully better as they get richer. These authors tried 100,000 different combinations of indicators and weights, specifically looking for a way to break the relationship. They couldn’t find one.

The share of income spent on necessities declines as income rises. This means that necessities are consumed intensively at low-income levels, and the income-expansion path for basics is steep at first, then flattens out. That is precisely what the non-linearity in these data shows. The strongest effects of growth on basics are at the lowest income levels, where growth moves people from near-nothing to something. Once a country’s GDP per capita reaches around $20,000–$25,000, the marginal gains to basics from further growth are small, not because growth stops mattering, but because the basics have already largely been achieved.

For the countries where development economics actually operates – countries below the 80th percentile of global income – the typical person lives right at the peak of the elasticity curve. Growth there does not just raise measured income. It raises child survival rates, nutrition, schooling, and life expectancy. No targeted program does that at scale or as a path to universally decent living standards. Moreover – and importantly – the absence of growth ends any improvement.

The Policy Implication

Pritchett and Lewis are careful not to argue against targeted programs as such. A well-designed intervention that passes a benefit-cost test is worth doing. But there is a crucial distinction between “cost-effective” and “sufficient.” Targeted programs can be cost-effective at the margin without ever being sufficient to deliver universally decent living standards. Growth, if it actually happens, is both sufficient. That is the finding.

This distinction matters for how donors, governments, and international institutions allocate their scarce attention. If the goal is to get poor countries to rich-country levels of basic human wellbeing, the evidence says that the path runs through sustained GDP growth. The institutions, the property rights, the rule of law, the openness to trade and investment, the macroeconomic stability – basically the things Adam Smith advocated for 250 years ago – these are the things that produce growth, and growth is what produces the outcomes everyone claims to care about.

Development economists have spent two decades shifting their field’s attention away from the importance of growth. Instead, they committed themselves to the idea that poverty is a problem that can be solved with the right detailed programs. Pritchett and Lewis essentially argue that this reallocation of intellectual effort has been a major mistake.

Why This Matters Here at Home

On his Substack, Lant Pritchett writes about his new paper: “This paper is part of my continued quest to get development economics and development actors re-focused on economic growth, and yes, the more inclusive the better, but “inclusive” is just the adjectival modifier, “growth” is the noun. You might like having a red car more than a blue car but if you don’t have a car its color is hypothetical.”

The path out of poverty clearly runs through growth; nothing delivers results like it. Targeted programs are not remotely equal to growth, no matter what many development economists and activists claim. I am pretty certain this is also an indictment of pretending we can lift people out of poverty through the forced redistribution of a deliberately shrunk pie a la Piketty.

What We Measured

The Index is organized into three subindices, each capturing a different dimension of philanthropic policy.

The first, the Broad Policy Environment, examines the general economic conditions that power private giving. Philanthropy doesn’t emerge from a vacuum; it depends on a thriving private economy. When states pile on taxes, licensing burdens, and regulatory complexity, they shrink the pool of private capital available for charitable purposes. Prior research has consistently shown that people in economically freer states are more philanthropic. Our index takes that finding seriously.

The second subindex, Nonprofit Freedom, drills into the regulations specific to nonprofits: how easy is it to start a charity? What reporting requirements must organizations meet? How are paid solicitors regulated, and what compliance burdens do charities face? Standard economics predicts that higher barriers to entry and operation mean less charitable activity, and that dollars spent on compliance are dollars not spent on mission.

The third subindex, Donor Confidence, examines the rules that shape donors’ decisions: How does the state tax charitable contributions? Does it protect donor privacy? Can donors trust that their intent will be honored? How are donor-advised funds regulated? These factors affect not just whether people give, but how much, and how confidently.

What We Found

The differences between states are striking. Montana, Wyoming, South Dakota, Iowa, and Indiana rank as the five most philanthropy-friendly states in 2025. They combine economic openness, streamlined charity regulations, and strong protections for donor rights. At the other end, Connecticut, New York, Washington, New Jersey, and California impose the heaviest burdens, complex regulatory frameworks, weaker donor safeguards, and structural disincentives to giving.

The practical consequences are visible in the data. We measured charitable activity by counting the number of nonprofit organizations per billion dollars of GDP—a metric that adjusts for differences in state economic size.

The results are telling: among the five lowest-ranked states, there are an average of 63 charities per billion dollars of GDP. Among the top five, there are 122. That’s nearly twice as many charities per unit of economic output.

A one-point improvement in the Free to Give Index, roughly the difference between Nebraska and South Dakota, is associated with seven additional charities per billion dollars of GDP. While this analysis doesn’t establish causation, it’s a robust correlation that echoes earlier research: burdensome regulation suppresses the formation and growth of charitable organizations.

Why This Matters

Philanthropy flourishes when citizens are free to organize, give, and serve their communities without unnecessary government interference. Yet in too many states, charities and donors face layers of red tape that make it harder to form new organizations, raise funds, and direct resources where they are needed most. Every hour spent navigating duplicative filings, costly registration rules, or intrusive disclosure requirements is time and money diverted away from feeding families, housing the homeless, educating children, and strengthening civil society.

The Free to Give Index is designed to make those barriers visible. By benchmarking every state on the same consistent set of indicators, we give policymakers, researchers, and philanthropic leaders a tool to ask sharper questions: Is our regulatory framework protecting the freedom to give, or discouraging charitable entrepreneurship? Are state policies empowering donors and nonprofits, or imposing costs that ultimately weaken the communities they are meant to serve?

This is the inaugural edition of the Index, establishing a 2025 baseline. Future editions will track changes over time, allowing us to measure the effects of specific policy reforms as they unfold. As states experiment, loosening licensing requirements, reforming solicitation rules, strengthening donor intent protections, we’ll be watching and measuring.

America’s charitable tradition is a national asset. The Free to Give report is our effort to protect and strengthen it, one data point at a time.

There is a pattern that’s worth naming. Over the past several years, three French economists — Thomas Piketty, Emmanuel Saez, and Gabriel Zucman — have together produced a body of work that functions less like independent scholarship and more like a coordinated legislative program. Understand the logic of each piece, and the larger design becomes visible: a vast wealth-tax net drawn tight enough that no one can escape it, not by moving to Nevada, not by renouncing citizenship, not by relocating a business. Global, compulsory, enforced. For what goal? To level countries down.

They have told you that this is the goal. Believe them.

Sacramento First

Building on past work that they have done with Thomas Piketty, Emmanuel Saez and Gabriel Zucman, both Berkeley economists and longtime collaborators, co-authored a recent NBER working paper with Jasper Boll documenting California billionaire wealth and making the affirmative case for the state’s proposed one-time 5% wealth tax. This particular paper is a response to the Hoover Institution’s Joshua Rauh and coauthors showing that earlier revenue projections by Saez for the wealth tax are, well, bunk.

It’s worth noting that Piketty, Saez, and Zucman have been repeatedly caught by economists across the political spectrum, including Obama’s Treasury Secretary Larry Summers, inflating wealth-concentration figures, using nonstandard methods to manufacture desired, errors, and in at least one case quietly scrubbing prior data from the internet when new numbers told a more convenient story. Their work is less a research program than policy advocacy dressed up in academic clothing.

In this case, their paper argues that the California billionaire wealth tax could raise around $100 billion, and that mobility responses would be manageable. But Josh Rauh has run the same numbers and arrived at a sharply different conclusion. Jack Salmon and I have written about this matter in the past, but here is the basic problem with the Saez-Zucman projections: they counted billionaires who had already left. Larry Ellison departed California in 2020. Larry Page and Sergey Brin left before the proposed liability date of January 1, 2026. These departures were public record. Yet the Berkeley paper’s $100 billion figure does not adequately account for a tax base that was already walking out the door before the vote was cast.

Rauh’s team found that those confirmed departures alone reduce projected revenues by nearly 40% before a single dollar is collected. Apply the mobility elasticities that the academic literature actually supports. Factor in the future California income taxes permanently foregone from departed taxpayers, and the tax produces a net present value loss to the state of at least $25 billion.

These economists should know better, because the French government has already run this experiment. Encouraged by Piketty and Zucman, France tried to tax the very rich. From Fortune: from 2000 to 2017, around 60,000 millionaires chose to leave the country. The revenue collection took a hit. France largely repealed its wealth tax in 2018.

The pattern repeats itself in Washington state, where the story is unfolding in real time. That state’s new 9.9% millionaire’s tax does not take effect until 2028, yet it has already triggered a wave of departures. Rich Barton, co-founder of Zillow, recently announced he is now a resident of Las Vegas. In 2023, Jeff Bezos left for Miami. Howard Schultz decamped to Florida. Starbucks has been expanding its corporate footprint in Tennessee. Former Starbucks CEO Schultz wrote that his decision to leave “had much to do with family choices and my stage of life,” and then spent the rest of his op-ed explaining exactly how hostile Washington’s political class has become to business.

Governor Bob Ferguson’s own Office of Financial Management has now sent a letter to state agencies acknowledging “significant budget shortfalls” and warning them explicitly: do not assume the millionaire’s tax revenue will materialize, because the law faces a court challenge, potential ballot measures, and deep uncertainty.

But what the disagreement obscures is something more basic: even granting Saez and Zucman their best-case revenue scenario, the California wealth-tax’s own logic predicts the problem it cannot solve. Billionaires are mobile. The wealthier they are, the more mobile. California cannot wall them in. And Saez and Zucman know this perfectly well.

Which is why Zucman has spent years building the cage.

The Global Backstop

In June 2024, Zucman published a blueprint commissioned by the Brazilian G20 presidency: a coordinated international minimum tax on billionaires, set at 2% of wealth annually, designed to raise $200–250 billion per year from roughly 3,000 taxpayers globally. The pretense is that these guys aren’t paying their fair share. He gets there, as he does in the case of California, by making questionable claims. But at the very least he is clear about the point of this effort. Prevent high income earners from voting with their feet in response to local, state and federal wealth and income taxes.

His own words: “A common challenge with increasing the taxation of ultra-high-net-worth individuals is the international mobility of the taxpayers involved.” The entire architecture of international coordination, the information exchange requirements, the “undertaxed payments rules” modeled on the OECD’s corporate minimum tax, the provisions for taxing former residents of participating countries for years after they depart, exists for one reason: to ensure there is nowhere left to go.

This latest paper even contemplates taxing billionaires based not on where they live, but on where their companies have assets or employees. If a U.S. billionaire moves to Dubai, and his companies have operations in France and Germany, France and Germany would be authorized to collect a share of the wealth tax. His companies could be made liable if he refuses. This is not a revenue-raising exercise with some mobility concerns attached. It is a mobility-suppression exercise with revenue attached.

The sequencing matters. Saez and Zucman push the California wealth tax knowing full well that billionaires will leave. Zucman simultaneously builds the intellectual and policy case for why that flight is exactly the problem requiring an international solution. The California tax creates the political demand for the federal response; the federal response creates the demand for the international one. Each step makes the next step feel necessary.

And Then There Is Piketty

Thomas Piketty’s recent contribution is the most revealing and the least encumbered by pretense. All these wealth and income taxes proposed over the years are really meant to achieve a degrowth agenda and make everyone poorer. That’s the plan.

Don’t take my word for it. Look at the comprehensive global economic restructuring program released in June 2026 by Piketty and his team. It is presented as designed to kill two birds with one stone: fix the climate and end global inequality, but what it is really doing is establishing some world governmental organizing aimed at making rich countries poorer. It would do that by capping their GDP per capita at roughly $69,000 (the US is currently at $94k), limit global economic growth to between 0 and 0.5% annually, mandate a three-day work week, reduce construction activity by 70%, manufacturing by 87%, and fund a “Global Justice Fund” through international wealth and income taxes administered by a new supranational bureaucracy answerable to no existing democratic institution.

This is actually a description of what he is proposing.

Piketty assumes vast stocks of billionaire wealth available for redistribution while simultaneously projecting near-zero long-term growth. But wealth is not a static pile. It is the capitalized present value of expected future returns. Destroy growth, and you destroy the very asset base you planned to tax.

There is a second contradiction. The plan is supposed to lift up sub-Saharan Africa. How does a region grow at 4% annually once the wealthy countries and companies currently buying its exports and financing its development have been deliberately contracted? Through the Global Justice Fund, you may say, the same fund that was depleted because the degrowth plan destroyed the value of the assets that were supposed to fund it.

Worse still is the climate baseline. It relies on a 4.5 percent warming scenario. It is very close to the RCP 8.5 scenario, warming of approximately 4.8°C by 2100, which the scientists who built that scenario officially retired it in May 2026 as implausible, with the updated central projection closer to 3°C. Recently Piketty responded to critics like me on X who pointed out that fact by saying his team is not using the RCP8.5 scenario but their own modeling framework. But that’s hardly an argument for why it is justified to continue using such alarming numbers when scientists themselves are dropping them.

This proposal is not just happening on the tail of excellent news about climate; it is happening as “global inequality is at his lowest level in nearly 150 years”, points out Felix Salmon. It is also happening at the time when we have much empirical evidence that the best way to lift people out of poverty is growth, and, that economic prosperity is also really good for the environment. Also, economic growth is compatible with declining resources use.

On the other hand, phil Magness is correct that, “”Degrowth Economics” is the exact type of policy regime that will lead to poverty, famine, malnutrition, disease, and an immiserating decline in worldwide living standards.”

Let’s sum this up: Here we have a plan to force the global economy into managed contraction, making rich countries deliberately poorer, justified by a climate scenario its own authors’ preferred scientific authority has disavowed, at the time when global inequality is the lowest it’s been. When all justifications disappear you are left with what is really the point of this exercise: Making rich countries poor, in part with wealth and income taxes and big government. That’s it. That’s the plan. It is the plan in California. It is the plan of those arguing for the federal wealth tax in the US, France and elsewhere. It is the plan of those arguing for a global wealth tax.

It is the plan and they are open about it. Please believe them.

P.S.: I failed to mention that Saez and Zucman are part of the World Inequality Lab, which published the collective grand plan that Piketty is a spokesperson for. So when you hear them talk about wealth tax in CA or in the U.S., keep in mind what the ultimate grand plan is about.

Nonetheless, many of his tariffs have gone through, and the direct impact on consumers’ wallets has been noticeably painful. According to estimates from Yale University’s Budget Lab, the current tariff regime will ultimately cost the average household up to $940 per year. But that is assuming the tariffs are allowed to expire, if the section 122 temporary tariffs are extended the cost could be up to $1500. The Budget Lab also estimates that the tariffs will leave the U.S. economy persistently smaller in the long run. When reading the June 1st proclamation, there is evidence that even the Trump administration itself is aware of the distortionary and negative effects its tariffs are generating.

Even though the administration publicly argues that tariffs are necessary for economic growth, it quietly creates new carveouts to undo its own mistakes. When discussing these tariff adjustments, the Secretary of Commerce cited a few key industries to target. He claimed, “recent circumstances have affected and are affecting domestic industries that use agricultural equipment, industrial equipment and machinery, and other related products.” In this case “recent circumstances” would seem to refer to the very tariff policies imposed by the administration.

The Secretary later recommends the administration should “expand the category of derivative products subject to the temporarily-reduced 15 percent ad valorem duty to include agricultural equipment and certain heating, ventilation, and air conditioning (HVAC) systems.” This is just another targeted carveout that tries to undo the current price distortions and limit the downstream effects these tariffs will create for these industries. Simply put, they are trying to shield certain industries from the costs of their tariffs, and in doing so, they implicitly admit the price distortions caused by their current regime.

Later in the proclamation, the White House acknowledged that many of these products they are targeting “serve an important role in productive domestic economic activity.” In other words, they understand that these are important industries, and they also understand tariffs have adversely affected them. Yet instead of repealing these destructive tariffs, they quietly try to amend them back toward freer trade. If tariffs were working as advertised, there would be little need to keep shielding the very industries they are supposed to help. This is the fundamental problem with tariffs: they aim to protect domestic industry, but they then require carveouts to avoid damage to the very industries they try to protect, all while they raise prices for consumers and producers.

As the administration continues to amend these distortionary tariffs, it is key that we keep pointing out the real message they are sending. Unjustified tariffs are simply taxes, and they needlessly hurt both domestic consumers and industry. So while the June 1 proclamation might seem like a simple policy amendment, it really serves as a quiet reminder that even the White House must face the reality of the economic hardships caused by tariffs.

This a guest post by Cameron Ewine.

Now, a new paper in Structural Change and Economic Dynamics by Justin Callais, Vincent Geloso, and Alicia Plemmons pushes this inquiry further. Rather than asking whether capitalism produces bad environmental outcomes in the abstract, the authors ask a sharper question: What actually happens to environmental outcomes when countries undergo major episodes of economic liberalization? The results challenge one of the most persistent assumptions in environmental politics.

From Theory to Episodes

The authors study 49 episodes of major economic liberalization since 1970. These are not marginal policy tweaks. These are countries that substantially opened trade, strengthened property rights, reduced regulatory burdens, and increased economic freedom in ways that are measurable and historically significant.

The main analysis uses matching methods to construct appropriate comparison groups, and the authors replicate key results using synthetic control methods on the most extreme liberalizers.

The authors find that GDP per capita rose roughly 16% within ten years of major liberalization episodes. But environmental outcomes did not deteriorate alongside that growth.

There was no increase in total greenhouse gas emissions. Death rates from air pollution fell. In many post-2000 cases, emissions per capita and emissions per dollar of output declined. The authors find these results consistently, across different specifications and different subsets of countries.

This matters because the standard narrative in much of environmental policy is that growth and environmental protection are in fundamental tension. More output means more resource use, more emissions, more pollution. Under this view, economic liberalization, which tends to accelerate growth, should worsen environmental outcomes. The data from 49 real-world episodes says otherwise.

Scale Is Not the Whole Story

The reason for that standard narrative is not entirely wrong—it focuses on what economists call the scale effect. More production can mean more resource use. That’s a real force.

But it’s only one of several forces at work. As countries grow richer, they tend to shift toward less resource-intensive sectors. They innovate, producing more with fewer inputs. And their citizens demand higher environmental quality. Taken together, these forces generate what economists call the Environmental Kuznets Curve: pollution may initially rise during early development but eventually peaks and declines.

The crucial question, and this is where Callais, Geloso, and Plemmons make their most important contribution, is not whether such a curve exists, but what determines where that peak falls. Their argument is that institutions determine the answer.

Growth generated under secure property rights, open markets, and the rule of law tends to reach the turning point sooner and at lower levels of environmental degradation. The EKC’s peak arrives earlier and lower. That is a very different claim from simply saying “growth is fine.” It says that the type of growth matters, and that the institutions underpinning liberalization are themselves part of the environmental story.

Why Institutions Matter for the Environment

The mechanism is worth taking seriously. Property rights force polluters to internalize costs. Pollution often involves nuisance, trespass, negligence, and liability. Those who impose costs on others face compensation claims, legal action, higher insurance premiums, and reputational damage. The result is pressure to reduce environmental harms before regulators ever intervene.

The same institutions that protect property rights also encourage innovation and conservation more broadly. Entrepreneurs have incentives to develop technologies that use fewer resources. Market prices communicate scarcity and encourage substitution. Consumers can reward environmentally responsible firms and punish irresponsible ones. Self-interest, under the right institutional framework, becomes aligned with environmental stewardship.

The authors went a step further and decided to review the experience of de-liberalizing regimes. Using the same methodology, observing the de-liberalization period in Venezuela, the authors found that as the country grew poorer during de-liberalization, emissions per dollar increased significantly compared to the counterfactual.

This is not a claim that markets solve all environmental problems. The authors are careful on this point: externalities exist and matter. But the empirical record from these 49 liberalization episodes suggests that the institutional environment in which growth occurs shapes environmental outcomes in meaningful ways, and that liberalization, on net, has not been the environmental disaster many assume.

The Bigger Picture

Read alongside the earlier Geloso et al. paper on the doughnut model, a consistent picture is emerging from the empirical literature. The degrowth framework predicts that capitalism and economic freedom systematically worsen social and ecological outcomes. Those predictions don’t hold. And now, when we look directly at episodes of major liberalization—the closest thing to a natural experiment we have—the environmental picture is considerably more optimistic than the standard narrative suggests.

This does not mean growth solves everything, or that environmental policy is unnecessary. But it does mean that the framing of a fundamental choice between prosperity and preservation is empirically unsupported. Countries can grow richer and achieve better environmental outcomes. The institutions that enable one tend to enable the other.

That is a finding worth sitting with, especially for those who have built policy frameworks on the assumption that the opposite is true.

Paragon estimates that roughly 6.2 million exchange sign-ups in 2026 were improper, representing about 27 percent of all ACA exchange sign-ups. The associated taxpayer cost could be as high as $25 billion this year. This is a predictable consequence of a program designed around excessive subsidies, weak verification, and third-party actors who are rewarded for maximizing enrollment rather than ensuring eligibility.

The core problem is simple. When government makes something “free” at the point of purchase, it does not eliminate cost. It changes who pays, who decides, and who has an incentive to police abuse. In the ACA exchanges, generous subsidies flow directly to insurers on behalf of enrollees. Many plans require little or no premium from the consumer. Brokers and enrollment intermediaries can receive commissions for placing people into plans. And if the applicant’s reported income qualifies them for the maximum subsidy, the transaction can be completed with minimal resistance.

Paragon’s methodology focuses on the 100 to 150 percent of the federal poverty level income category, where ACA subsidies are most generous. The authors compare the number of people enrolled in that category with Census-based estimates of the plausibly eligible population. In many states, enrollment substantially exceeds the number of people who could reasonably be eligible. The excess is treated as improper enrollment.

That approach is not perfect, and Paragon is right to describe its estimates as estimates. But the scale of the mismatch is too large to wave away. In Florida, the paper estimates that improper enrollment is especially severe. Texas, Georgia, South Carolina, and North Carolina also account for a large share of the national problem. The pattern is not random. It is concentrated in states using HealthCare.gov, particularly non-expansion states where low-income individuals have a strong incentive to report income above the poverty line in order to qualify for exchange subsidies.

This is what economists mean when we talk about incentives. Policy does not merely transfer money; it creates margins of behavior.

The ACA created a cliff. People below 100 percent of the poverty line in non-expansion states generally do not qualify for exchange subsidies. People just above that threshold can qualify for large subsidies. Once policymakers draw that line, they should not be surprised when reported income clusters on the favorable side of it. Nor should they be surprised when brokers, insurers, and enrollment platforms discover that the most profitable customer is one whose premium is paid by Washington.

The problem is compounded by zero-premium plans. A plan with no monthly payment is much easier to sell, much easier to retain, and much easier to manipulate. If someone is enrolled without fully understanding what happened, there may be no monthly bill to alert them. If a broker switches a consumer into another plan, the consumer may not notice if the premium remains zero. If an enrollee never uses the plan, federal subsidy dollars can still flow.

Paragon points to evidence of “phantom enrollment,” including large numbers of exchange enrollees with no medical claims. Again, no single statistic proves fraud in every case. Some people are healthy, while some people enroll and simply do not need care. But when millions of people are enrolled in heavily subsidized coverage, many pay nothing, many have no claims, and enrollment exceeds plausible eligibility counts, the burden of proof should shift to those defending the status quo.

This is a recurring theme in modern progressive governance. A program is created to solve a real problem. Subsidies are expanded to make the program appear more generous. Administrative safeguards are relaxed to maximize take-up. Then, when predictable abuse follows, critics are told that concern over fraud is merely hostility to the poor.

But program integrity is not hostility to the poor. It is respect for taxpayers, for genuine beneficiaries, and for the rule of law.

Every dollar spent on improper enrollment is a dollar not available for deficit reduction, lower taxes, or better-targeted assistance. Worse, fraud and phantom enrollment corrupt the information policymakers use to evaluate the program. If enrollment numbers are inflated by people who are ineligible, unaware, duplicated, or not actually using coverage, then the headline enrollment statistics become propaganda rather than evidence.

The lesson here is that open-ended federal subsidies, routed through private intermediaries and enforced by weak verification, will predictably invite abuse. Washington tried to engineer affordability through third-party payment and bureaucratic formulas. The result was not discipline, transparency, or consumer control. It was a system in which insurers collect subsidies, brokers chase commissions, and taxpayers are left holding the bag.

Paragon recommends stronger eligibility verification, tighter identity authentication, stricter oversight of enrollment intermediaries, limits on automatic reenrollment, and reforms to zero-premium plan structures. Those are all sensible steps, but the deeper reform is to stop pretending that subsidized enrollment is the same thing as real coverage, and that federal spending is the same thing as health care.

Obamacare’s fraud problem is not an accident. It is the result of policy design. When government subsidizes something too generously, verifies eligibility too weakly, and measures success by how many people are signed up, it should expect exactly what Paragon finds: millions of questionable enrollments and billions of taxpayer dollars at risk.

It’s what the Department of Government Efficiency (DOGE) was founded on. It’s a major source of savings in budget resolutions. And the president has ordered agencies to stop it and created a new task force to eliminate it.

But little has been accomplished to show for it.

The reason is straightforward: Waste, fraud and abuse are difficult categories to define, challenging to measure and costly for politicians to monitor.

In other words, the phrase “waste, fraud and abuse” is politically poignant but effectively meaningless. Until better measurements become available, policymakers should retire this popular catchphrase and focus their attention exclusively on improper payments. Doing so would replace vague promises with measurable accountability for how taxpayer dollars are spent.

The Problems with “Waste, Fraud, and Abuse”

Beginning first with fraud: Fraud is relatively well-defined as “willful misrepresentation,” but it is difficult to measure in a way that is useful for ongoing oversight. Looking at fiscal years (FY) 2018–22, the Government Accountability Office (GAO) estimated that the government incurred annual losses of between $233 billion and $521 billion to fraud.

But as GAO explained in that report, this estimate is not predictive. The actual level of future fraud depends on a variety of factors including future threats and new emergency spending. It also overlaps with the COVID years, when emergency programs experienced substantial fraud.

Unfortunately, as GAO notes, “It is not possible to break out a subset of our government-wide estimate to describe pandemic program fraud.” Consequently, it is unlikely that GAO’s estimated fraud range represents either pre- or post-COVID norms.

The inherently subjective nature of waste and abuse makes them even more problematic metrics than fraud. That does not mean questionable spending never exists; it means there is no neutral or consistently measurable standard for identifying it across the federal government. Further, spending that appears wasteful to taxpayers broadly is not wasteful to the beneficiaries of that spending.

For every obvious example of abuse, such as a $10,000 toilet seat cover, there are numerous high-cost contracts that require judgements Congress is poorly positioned to make about quality, necessity and tradeoffs. Other seemingly clear-cut examples of waste—such as $240,000 in funding for a Pakistani cartoon climate series or $375,000 for a dance festival—exist because there are government officials, including politicians, who choose to fund them.

This subjectivity makes impartial and consistent measurement of waste and abuse effectively impossible. Combined with its lack of reliable estimates of fraud, Congress is left with little reason to rely on the phrase “waste, fraud and abuse.” This is where improper payments become handy.

What Does Improper Payment Data Tell Us?

In contrast to the limitations of “waste, fraud and abuse,” improper payments are objectively measurable, breakable into categories and reported regularly—providing a useful operational metric for policymakers. Improper payments include the following:

• overpayments and underpayments, which are payments made in the wrong amount or to the wrong person

• technically improper payments, which are those that fail to meet statutory or regulatory requirements even if they are for the right amount and to the right person

• unknown payments, when an agency cannot determine whether payment was proper or not

Federal agencies estimated $185.8 billion in improper payments in FY 2025, with roughly 82% attributable to overpayments. Since FY 2004, estimated improper payments have totaled $3 trillion, as shown in Figure 1.

Overpayments have comprised the vast majority of improper payments. Between FY 2015 and FY 2018, overpayments accounted for 92.3% of all improper payments, and since FY 2021 they have averaged 82.4%. The exceptions are FY 2019 and FY 2020, when agencies first began recording improper payments as “unknown” and “technically improper.” This breakdown is shown in Figure 2, starting in FY 2015.

The improper payments data also points to where payment integrity breaks down. Starting in FY 2021, Office of Management and Budget guidance instructed agencies to categorize improper payments into one of the following root causes:

1. when the required data or information to verify eligibility do not exist

2. the data exist but are not accessible to the agency or payor

3. the data exist and are accessible, but the agency or payor fails to access them

4. insufficient documentation, such as from states, vendors, or applicants

5. technically improper payments

These root causes are illustrated in Figure 3.

Since 2021, the largest cause of improper payments has been failure to access the data or information necessary to verify eligibility, averaging $142 billion in improper payments per year, or $712 billion total over five years, equal to 64% of all improper payments.

The second largest cause of improper payments has been inability to access data needed to verify eligibility, at $27.4 billion per year ($137 billion total or 12% of all improper payments).

The third and fourth largest causes were substantially smaller: data not existing at $23.9 billion per year ($120 billion total or 11%) and insufficient documentation at $21.7 billion per year ($109 billion total or 10%), respectively. Technically improper payments averaged just $6.8 billion per year ($33.9 billion or 3%).

Failure to access data plays an even greater role when looking just at overpayments, as visualized in Figure 4. In FY 2025, 74% of all overpayments resulted from a failure to access the necessary data to verify payment accuracy, compared to just 8% caused by an inability to access the necessary data. Since FY 2021, failure to access data has accounted for $671 billion in overpayments, equal to 73% of all overpayments and 60.3% of all improper payments.

Overpayments can be further classified by whether the payment was “within agency control” or “outside agency control”—that is, whether it was within the agency’s power to prevent the improper payment or not. In FY 2025, 93% of all overpayments were categorized as being outside of agency control compared to just 7% within agency control.

And of course, we can categorize improper payments by federal program, as in Figure 5. The bulk of improper payments in FY 2025, $126 billion or 68%, were concentrated in just a handful of programs: Medicare, Medicaid, the Earned Income Tax Credit and SNAP. The remaining $60 billion were spread across 58 other programs.

Conclusion

In total, improper payment estimates tell us that:

1. improper payments are substantial and rising

2. most improper payments are overpayments

3. they are mostly due to a failure to access existing and accessible data or information

4. they are almost entirely outside of agency control

5. they are concentrated in a handful of programs

Congress should spend less time promising to eliminate “waste, fraud and abuse” and more time focusing on the payment errors agencies already estimate, categorize and report. Improper payments are not a perfect measure, and they should not be treated as the same thing as fraud or guaranteed savings. But they offer a better starting point than the usual catchphrase. They are measurable, concentrated in a small number of programs and often tied to a failure to use information that already exists. That makes improper payments a more useful target for oversight than a phrase that is politically useful precisely because it can mean almost anything.

In subsequent posts I will discuss the limitations of improper payment estimates, explore additional information we can gather beyond the aggregate data, and attempt to assign responsibility over improper payments.

Unfortunately, the paper entirely sidesteps the question it should be answering.

The California Billionaire Tax Act does not justify itself on abstract grounds of tax fairness. It justifies itself on a specific fiscal claim: that the One Big Beautiful Bill Act created a roughly $19 billion annual budget hole in California’s healthcare safety net that requires emergency revenue to fill. That is the premise on which the initiative was drafted, campaigned for, and is being sold to voters.

A new Hoover Institution analysis released this week by Joshua Rauh, Tom Church, Daniel Heil, Benjamin Jaros, and John Doran examines that premise in detail. What they find should be deeply uncomfortable for anyone relying on the Saez-Zucman framing.

The $19 Billion Figure Is a Fabrication of Averaging

The proponents’ $19 billion annual shortfall figure is not an estimate of any actual annual loss. It is a cumulative ten-year federal Medicaid reduction divided by ten to produce a tidy annualized average. The Hoover paper exposes how misleading this framing is.

Using section-by-section CBO cost estimates, Rauh et al. find that the OBBBA will reduce federal Medi-Cal spending by $156 billion over ten years, a figure consistent with Kaiser Family Foundation estimates. But only 35% of those reductions occur through 2030. The annual reduction in 2026 is roughly $2 billion. It does not reach $19 billion in any single year until the early 2030s. Presenting a back-loaded worst-case scenario as an immediate annual emergency is not analysis; it is advocacy dressed up in decimal points.

Even the $156 billion headline does not translate into a $156 billion state budget problem. The Hoover paper carefully disaggregates who actually bears these reductions. The eligibility provisions, work, school, and volunteer requirements, account for $89 billion of the total. These provisions reduce Medi-Cal enrollment and therefore reduce both federal and state spending.

The state actually saves approximately $15 billion from these provisions over the budget window. The remaining $62 billion comes from limits on provider taxes and state-directed payments, which creates a funding gap that either the state backfills or providers absorb. The state is not automatically on the hook for the full $156 billion, and proponents who cite that figure as a “budget hole” are being deliberately imprecise.

The Tax Is Structurally Mismatched to the Obligation

Here is the deeper problem that the Saez-Zucman paper cannot paper over with wealth statistics: even if you accept the proponents’ revenue estimate of $100 billion at face value, the Billionaire Tax Act is the wrong instrument for this obligation.

The tax collects revenue over five years. Two-thirds of the OBBBA’s Medi-Cal reductions fall after 2030. The money runs out before the problem it supposedly addresses fully materializes. The Hoover paper estimates that after netting out lost income tax revenue from departing billionaires, the $40 billion in realistic collections is exhausted by 2029, the very moment the annual federal reductions start accelerating past $15 billion per year and growing.

California also has a $93 billion structural deficit projected through 2030, and a well-documented institutional habit of redirecting earmarked Medi-Cal funds toward general budget priorities. The Legislative Analyst’s Office (LAO) has noted this pattern explicitly with respect to excess MCO provider tax revenues.

Structural deficits are not new in California. The LAO notes that the legislature faced a $27 billion deficit in 2023-24, a $55 billion deficit in 2024-25, and a $15 billion deficit in 2025-26.

There is no mechanism in the Billionaire Tax Act that would prevent $100 billion in one-time wealth tax collections from disappearing into that deficit well before the largest federal reductions arrive. Saez and Zucman do not address this.

Medi-Cal’s Problem Is Spending Growth, Not Insufficient Revenue

The Boll-Saez-Zucman paper rests on the implicit assumption that California’s healthcare safety net is underfunded and that additional revenue would address the problem. The Hoover analysis demolishes this premise.

Medi-Cal expenditures have nearly tripled since 2011, growing from $54 billion to $185 billion annually. Over that same 15-year period, the Department of Health Care Services’ General Fund share grew at a compound annual rate of 9%, compared to 5.8% for all other General Fund spending. Medi-Cal is now the second-largest General Fund program in the state, comprising 40% of spending. If its General Fund share had simply been held at the post-ACA, pre-COVID average of 15.7%, the state would save roughly $9.8 billion per year, nearly $88 billion over ten years.

The explosion in Medi-Cal spending is the result of changes in the law that have massively increased the population covered by the program. There were around 12 million on Medi-Cal in 2015, and about 15 million today, or nearly 4-in-10 Californians. The LAO also suggests that per-enrollee spending can be explained by major policy changes enacted since 2017-18, such as the expansions of comprehensive coverage to undocumented adults.

And what did California get for this extraordinary spending growth? Quality of care has remained below the national median across virtually every measured dimension for both adults and children, consistently and without interruption. The Hoover paper documents this through five rounds of Consumer Assessment of Healthcare Providers and Systems survey data. In 2024, Medi-Cal ranked below the 50th national percentile on all eight quality metrics for both adults and children. The program’s challenges are structural, not fiscal. Pouring more money into a poorly designed system does not fix the design.

California Has the Tools. It Chooses Not to Use Them.

Perhaps the most important finding in the Hoover paper is that California does not actually need new revenue to address the gap that would remain after accepting OBBBA’s eligibility defaults. If the state accepts the federal work and community engagement requirements, which represent the default condition of federal law, not some radical policy imposition, the remaining federal reductions the state would need to address fall to approximately $67 billion over ten years.

The Hoover analysis identifies policy levers already within the state’s control that generate over $125 billion in ten-year savings without any new revenue source: reversing discretionary Medi-Cal coverage expansions for undocumented immigrants ($37 billion over ten years) and returning the program’s General Fund share to its post-ACA, pre-COVID average ($88 billion over ten years). That is nearly double the remaining obligation.

None of this requires taxing anyone. It requires California to make different spending choices than it has made over the past decade. The Saez-Zucman paper does not engage with this alternative because doing so would require acknowledging that the “fiscal emergency” framing is, at minimum, incomplete.

The Real Problem with the Proposal

Boll, Saez, and Zucman have written a paper about billionaire wealth. It is not, despite appearances, a paper about whether the California Billionaire Tax Act makes fiscal sense. None of it is relevant to whether a one-time residency-triggered levy is a structurally sound way to address a back-loaded, partially discretionary reduction in federal Medicaid cost-sharing.

One of the paper’s authors helped design the ballot initiative. The paper’s framing reflects that involvement. When you begin with the conclusion that billionaires should pay more and work backward to the policy, you tend to produce analysis that answers the question you want to answer rather than the one the voters will actually be deciding.

The Hoover Institution paper asks the right question. Its answer is that the fiscal premise of the Billionaire Tax Act fails on its own terms, in magnitude, in timing, and in structure. That finding will not make it into the New York Times op-ed. But it should inform how Californians vote in November.

“For decades, the top tax rate on the wealthiest Americans — like @JeffBezos — was much higher than it is today.

What did we do with that money? We built roads and bridges, expanded healthcare, and invested in education. We grew the middle class.”

Yes, top tax rates were significantly higher in the past than they are today. This is about the only part of Steyer’s post that is accurate.

Today, in several states, the top earners pay combined federal and state marginal tax rates north of 50%, but for the sake of simplicity, let’s focus on federal income taxes over time. Then we can address Steyer’s other points that lead on from this one.

Let’s take 1952 for example, when the top income tax rate was 92 percent (compared to 37 percent today). Steyer asks what did we do with all that money? Well, first we need to observe how much “all that money” amounts to with a 92 percent top rate.

In 1952, the federal government raised 18 percent of GDP in tax revenues. This compares with federal revenues of 17 percent of GDP today. So, relative to today, “all that money” amounted to only about 1 percent of GDP in federal revenue, roughly $300 billion in today’s economy, or about 2 weeks of current government spending.

Steyer then argues that with “all that money”, we built roads and bridges, expanded healthcare, and invested in education.

Let’s start with roads and bridges. In the 1952 fiscal budget, Congress spent $1.69 billion on transportation and communication. In 2025 dollars, this is about $20.5 billion. By comparison in the FY 2025 federal budget, the federal government spent $145 billion on transportation.

In other words, when top tax rates were 92 percent, the federal government spent 7 times less on transportation than it does today.

What about health care and education? Well, the federal government provided tiny amounts of funding to healthcare in the 1950s, as this was before the creation of health care entitlements in 1965. As for education, about $5.8 billion in 2025 dollars was provided by the federal government in 1952, compared to $124 billion in the FY 2025 budget, or more than 20-times the amount.

I want to be fair to Steyer, so I will review a second federal budget where the top tax on the rich was still high by historical standards but includes post-1965 entitlement program spending. Let’s review the 1979 budget when top income tax rates were 70 percent.

Adjusting for inflation, in 1979, the federal government spent $222 billion on all health care programs. Compare this to the FY 2025 budget which allocated $1.66 trillion to Medicare and Medicaid combined, or more than 7-times the amount in 1979.

For transportation and education spending in 1979, the federal government spent roughly half as much on both as it does today, adjusting for inflation. Even if we adjust for population growth, the federal government still spends significantly more today on transportation, education, and especially on health care.

So, we have determined that “all that money” that exorbitantly high taxes on the rich raised amount to a measly $300 billion in today dollars, and we’ve also determined that Steyers claim that this money was used to fund transport, health care, and education doesn’t stand up to the scrutiny of the actual budget data.

Before moving onto Steyers final claim, it is worth noting that, even though top marginal tax rates on the rich were significantly higher than today, almost no one actually paid income taxes at those levels. This is exactly what the empirical literature tells us about how high earners adjust behavior in response to high tax burdens.

In 1952 when the top income tax rates were 92 percent, the top 1 percent paid an effective rate of less than 17 percent, while the top 0.1 percent paid a 21 percent effective income tax rate.

Today, both the top 1 percent and 0.1 percent pay effective income tax rates above 26 percent. So, with a top rate of 37 percent, the wealthiest Americans pay higher effective income tax rates than they did with a top rate of 92 percent.

Now, we can address Steyers final claim that follows from his priors: “We grew the middle class”.

Stephen Rose and Scott Winship at the American Enterprise Institute have done extensive work on measuring changes in the middle class over time as far back as 1979. The first notable datapoint they highlight is that median family income has risen by 39 percent since 1979, or 52 percent if we adjust for changes in family size over time.

Perhaps, more importantly, they find that, yes, the share of the population that is core middle class has shrunk over time, but this entire shrinking pattern reflects net upward movement. While fewer and fewer families are poor or middle class today, it is because they have moved up to upper middle class.

Contrary to Steyer’s argument about growing the middle class, most American families in 1979 were either poor or lower middle class, while today most Americans are core middle class or upper middle class.

All said and done, only the first sentence of Steyer’s X post is accurate. Taxes on the rich were notably higher in the past than they are today, but the government was also a lot less involved in transferring funds to transport, education, and health care among other income support and transfer programs. Perhaps most importantly, American families today enjoy substantially higher living standards than families did during the era of confiscatory marginal tax rates.

This makes it a good time to revisit past beliefs and policies.

For the better part of fifteen years, the dominant fiscal narrative in Washington and in many economics departments ran roughly like this: the federal government can borrow cheaply, the economy will grow faster than the interest rate on its debt, and therefore the fiscal cost of additional borrowing is small or even zero. That’s the famous r-g. This argument was deployed to justify the post-2008 stimulus, the pandemic response, and the spending expansions in between. Its most prominent advocates included former Treasury secretaries and the most cited economists of the era.

This argument was wrong, and not only the way forecasts are sometimes wrong. It was wrong in its foundations.

Rates were Always Going to Go up:

When the ten-year Treasury yield fell below one percent, many economists and policymakers concluded that the long-run fiscal constraints their predecessors had worried about had been substantially relaxed. Better yet, rates would always stay this low, demand for government bonds was supposedly insatiable, and inflation was a thing of the past. As such, the logical conclusion was that it was okay to take advantage of cheap money, invest in public goods, don’t worry about the debt.

That was always shaky grounds, in my opinion. Jack Salmon and I wrote about it here among other places. First, it seems doubtful that rates would always stay so low. There were a bunch of reasons for that. First, Social Security and Medicare carrying unfunded obligations in the range of $70 trillion over the relevant horizon. Then, every emergency in recent history has added 20 to 30 percentage points to the debt-to-GDP ratio without the implementation of austerity measures once the crisis is over. In other words, the government was certain to be a very large and increasing borrower for a very long time. To believe that surging borrowing demand will not drive up yields, one must assume that the supply of savings willing to absorb it at low rates is essentially infinite.

It is not. In fact, in 2020, foreign investors had also started to reduce their demand for Treasuries. There was plenty of evidence to suggest that the rate environment of the 2010s was not permanent.

Deficits Were on a Path Going Up, not Staying Constant

Second, the standard claim that when growth exceeds the interest rate, government debt to GDP decline is technically correct in a narrow case. If the interest rate remains below the growth rate and the deficit remains constant, the economy will eventually outgrow the debt. Fair enough.

But the math only works for a constant deficit. The United States does not have a constant deficit. Again, Social Security and Medicare spending rise year after year as the population ages and politicians hand out goodies to seniors like the OBBB did with the senior tax credit. The primary deficit is not a fixed addition to the debt stock each year; it is a growing one. And when the deficit grows, the economy does not outgrow the debt, even if the interest rate stays comfortably below the growth rate. The debt ratio grows in step with the deficit, indefinitely.

This distinction between a fixed deficit and a growing one is not a technicality. It is the difference between a fiscal problem that the economy eventually absorbs and one that it doesn’t. The United States has been on the wrong side of that distinction for at least two decades. The argument that low interest rates made the debt sustainable was being applied to a deficit path it was never designed to cover.

Try to exploit r-g and you change the expectations of future fiscal backings

The entire framework treats the gap between the interest rate and the growth rate as a stable feature of the economy, something fiscal policy can exploit without that gap changing.

What holds Treasury yields at any given level is the expectation that future governments will run surpluses sufficient to service the debt. Markets price bonds against that expectation. When expectations are strong, yields are low. The problem comes when policymakers observing low yields infer that markets are comfortable with much more borrowing. The thing is that new and sustained higher borrowing is itself evidence that the political system is less committed to future surpluses than markets had assumed. The yields supporting the policy are held up by the expectation that it is being disconfirmed. When that expectation is revised, so is r-g.

In other words, the gap between the interest rate and the growth rate is not a parameter of the economy that policy navigates around. It is a joint product of the loan market and the government’s budget constraint. Under a credible fiscal regime, the equilibrium interest rate reflects that credibility. Under an eroding one, it adjusts upward. The act of attempting to exploit cheap borrowing changes the thing being exploited.

The Path to Fiscal Stabilization:

Now is also a good reminder that the federal government does not have a revenue problem. Federal receipts as a share of GDP are near their historical average. There are tax expenditures worth closing on efficiency and moral grounds, and closing them would help. But the deficit is not the result of under taxation. It is the result of a spending trajectory driven almost entirely by two programs: Social Security and Medicare. The entire deterioration in the federal fiscal outlook over the next several decades is attributable to those two programs. Discretionary spending cuts, however aggressive, will not close the gap. Revenue increases will not close the gap. A wealth tax will not close the gap. Structural reform of Social Security and Medicare is the only path to fiscal stabilization.

It is. But the deeper point, which I failed to make explicitly in my syndicated column due to the lack of space, was perfectly summed up on X by Jonathan Hoenig: “government isn’t just inept as an investor—it’s immoral. Government has one role: protect individual rights not “maximize investment returns.”

He’s right. The problem is not merely that the government makes for a lousy investor. Government investing changes the moral relationship between risk, reward, and accountability.

In markets, investment is disciplined by consequences. Private investors deploy capital that belongs to them or that’s voluntarily entrusted to them. If they make bad bets, they lose money, reputation, clients, and sometimes even their careers. Prices communicate information. Profits reward value creation. Losses punish mistakes.

This discipline is central to what makes markets work. Government, however, investing operates under entirely different rules – rules that virtually eliminate discipline.

Government officials allocate resources extracted through taxation or borrowing backed by taxpayers. If projects fail, these officials rarely bear meaningful personal consequences. The losses are socialized. The incentives instead are political. Success is often measured not by financial returns but by press releases, ribbon cuttings, strategic narratives, or vague, untestable claims about resilience, competitiveness, and national greatness.

It is politics wearing the language of capital allocation. A good recent example is the expansion of the U.S. International Development Finance Corporation into maritime insurance and reinsurance for vessels transiting the Strait of Hormuz. The political goals of the administration are obvious, and whether that’s a good financial decision is secondary to why it was made in the first place. And again it’s using other people’s money to do it. Why should taxpayers be conscripted into financing these bets?

Another reason why there should be no such government investments is that they poison the economy and capital markets.

When Washington pours billions into favored industries – whether green energy, semiconductors, electric vehicles, or strategic manufacturing – the relevant question is not only: “Will these investments succeed?” It is also: “What opportunities are being displaced?”

Capital is finite. Talent is finite. Engineering expertise is finite. Every subsidized dollar changes incentives across the economy, leaving perfectly great projects biting the dust because they now have to compete for capital with the federal government investors.

Here’s an example. Remember the Department of Energy’s 1705 loan guarantee program? Yeah, that’s the one that extended $538 million in government guarantee to Solyndra and Abound Solar, and such companies. The debate around the program tends to focus on visible failures: the bankruptcies, the wasted taxpayer money, the embarrassing headlines.

But the larger distortion is harder to see.

Government support artificially lowers financing costs for politically preferred industries and firms. It signals to investors that some sectors enjoy implicit public backing while others do not. Capital then flows not toward the most productive uses, but toward activities most likely to attract political support. These unseen costs are real and significant.

How many entrepreneurs redirected their efforts toward subsidy-eligible technologies rather than unsubsidized alternatives? How many investors chased government-favored sectors because returns were effectively backstopped by taxpayers? How much private capital was diverted away from innovations consumers might have preferred but politicians did not?

We rarely know the answers to such questions because those investments never happen or they happen but they face higher borrowing costs ab=nd other distortions. We never really see the unseen victims of the government investor class. We don’t see the what could have been. In other words, the result of the government investor class is not merely occasional failure. It is a systematically distorted investment and economic landscape. And that’s immoral.

Oh and by the way, government “investments” don’t simply distort markets by redirecting capital toward politically favored industries. They also weaken the recipients themselves, dulling the competitive pressures that force firms to innovate, cut costs, and serve consumers better. See this older but always relevant piece by Chris Edwards.

The irony is that many former financiers now entering government once understood this reality. They recognized that prosperity comes from decentralized experimentation, not centralized direction. Yet increasingly, they seem to believe as U.S. government officials they can play markets investors. The assumption of super-intelligence underlying this desire is seductive: put successful businesspeople in charge and the government will behave more like a market. But institutions matter more than résumés. A brilliant investor inside government stops functioning as a market participant because the market’s disciplinary mechanisms disappear.

Hoenig is correct. The government’s central economic role is to protect property rights, enforce contracts, preserve the rule of law, and maintain the conditions under which decentralized market discovery can occur.

Much of this debate, however, proceeds at a highly rhetorical level. Politicians and commentators often speak in broad generalities about “bringing jobs back” or “protecting American industry,” while critics frequently respond with equally broad appeals to textbook economics. What is often missing is a careful examination of the large and growing empirical literature on the actual observed effects of tariffs.

Over the last three decades, economists have produced a substantial body of quantitative research examining how tariffs affect prices, trade flows, manufacturing employment, productivity, and economic output. This literature spans dozens of countries and trade episodes, including the Canada–U.S. Free Trade Agreement, NAFTA, trade liberalization in Latin America and Asia, antidumping duties, and the U.S.–China trade war (2018-2026). The methods used in these studies range from firm-level panel data and difference-in-differences designs to regional labor-market analysis and structural trade estimation.

This essay synthesizes much of that empirical literature. Rather than focusing on theory or ideological arguments, I compiled quantitative studies (published since 1997) that estimate statistically significant effects of tariffs or tariff liberalization on key economic outcomes. The goal is not to conduct a formal meta-analysis, but rather to systematically review the direction and consistency of the evidence across different variables and contexts.

To keep the exercise manageable, I focus primarily on four categories of outcomes: consumer prices, manufacturing employment, trade volumes, and productivity/output. I largely exclude purely theoretical papers, narrative essays, and studies without identifiable quantitative estimates. The result is a dataset composed primarily of empirical papers using measurable tariff changes and econometric identification strategies.

Several important caveats are worth emphasizing at the outset. First, some studies examine broad trade liberalization episodes, while others study narrow sector-specific tariffs or antidumping duties. Some focus on protected upstream industries, while others examine downstream manufacturers exposed to higher input costs. Second, the finding in some studies that tariffs preserve employment in a protected sector does not necessarily imply positive economy-wide effects. Many studies distinguish between localized gains and broader aggregate costs.

Still, despite these differences, broad patterns emerge from the literature. The bar charts below summarize the direction of statistically significant findings across 56 studies. While counting studies is not the same thing as measuring the precise magnitude of effects, the evidence nevertheless provides a useful overview of where the empirical literature has converged, and where important disagreements remain.

The purpose of the figure above is not to estimate a single “average” tariff effect. The studies included here examine different countries, time periods, tariff regimes, margins of adjustment, and outcome variables. Some estimate the effect of tariff reductions during broad liberalization episodes, such as NAFTA, CUSFTA, or India’s 1990s reforms. Others examine tariff increases, antidumping duties, retaliatory tariffs, or recent U.S. trade-war episodes. Some use firm- or plant-level data, while others rely on regional labor markets, product-level trade flows, or macroeconomic panels.

For that reason, this analysis should be read as a synthesis of the direction of statistically significant findings, not as a formal meta-analysis. A study is counted as finding a positive effect when higher tariffs are associated with an increase in the relevant variable: prices, manufacturing employment, trade volume, or output/productivity. A negative effect means higher tariffs are associated with a decline in that variable. This distinction is especially important because a “positive” tariff effect is not necessarily economically desirable. A positive effect on consumer prices, for example, means tariffs raise prices. A positive effect on employment means tariffs preserve or increase employment in the relevant sector.

The most unambiguous result is that tariffs raise prices. In the studies reviewed here, 19 find statistically significant positive effects of tariffs on prices, while none find significant negative effects. This includes evidence from antidumping duties, broad tariff increases, washing-machine tariffs, the 2018–19 U.S.–China trade war, and more recent tariff episodes. The price measures vary across studies, import prices, factory-gate prices, retail prices, CPI, and PCE goods prices, but the direction of the finding is strikingly consistent.

The magnitudes are often large as well. Blonigen and Haynes find that U.S. antidumping duties on Canadian iron and steel were passed through to import prices at more than the full statutory rate. Amiti and coauthors find that the Trump tariffs were largely borne by U.S. consumers and firms, with a 10 percent tariff increase associated with roughly a 10.4 percent increase in domestic prices. Flaaen, Hortaçsu, and Tintelnot find that washer tariffs raised washer prices by nearly 12 percent and even raised dryer prices, despite dryers not being directly tariffed. Cavallo, Gopinath, Neiman, and Tang similarly find high pass-through at the border, with more muted but still positive effects at retail. The basic lesson is straightforward: whatever their intended purpose, tariffs operate first as taxes on imports, and the empirical literature consistently finds that those taxes show up in prices.

The second clear result is that tariffs reduce trade volumes. Here, 12 studies find negative effects of tariffs on trade, while only one finds a positive effect. The negative findings appear both in studies of liberalization and protection. Clausing finds that Canadian goods receiving larger tariff reductions under CUSFTA saw much faster import growth into the United States. Romalis and Caliendo and Parro similarly find that NAFTA increased trade flows among member countries. On the protection side, Prusa finds that antidumping duties sharply reduced targeted imports, while studies of the recent U.S.–China trade war find substantial declines in targeted imports and exports.

The one positive finding in this category is not really evidence that tariffs increase trade in the ordinary sense. Rather, Fajgelbaum, Goldberg, Kennedy, Khandelwal, and Taglioni find that tariffs created export opportunities for third-party “bystander” countries. This is best understood as trade diversion. Tariffs suppress trade between the countries directly involved in the dispute, but some of that lost trade is rerouted through other countries or replaced by alternative suppliers. Thus, the overall trade-volume evidence is not that tariffs create more trade, but that they redirect and reduce targeted trade flows.

The employment evidence is more complicated and politically important. Among 12 relevant studies, 4 studies find positive effects of tariffs on manufacturing employment, while 8 find negative effects. Tariffs can preserve employment in protected industries. Revenga finds that tariff reductions in Mexican manufacturing were associated with employment declines in protected sectors. Trefler finds that the deepest Canadian tariff cuts under the Canada–U.S. Free Trade Agreement reduced employment by about 12 percent in the most affected industries. Dix-Carneiro and Kovak find that Brazilian regions exposed to larger tariff reductions suffered prolonged declines in formal employment and earnings. These studies imply that protection can maintain jobs in the industries or regions receiving it.

But this is only one side of the employment story. A larger set of studies finds that tariffs reduce employment once broader channels are considered. Flaaen and Pierce are especially important here because they explicitly separate the protection effect from the input-cost effect. They find that tariffs provided some benefit to protected industries, but this was more than offset by higher input costs and retaliation, producing a net decline in manufacturing employment. Waugh finds that goods-producing employment grew more slowly in counties more exposed to retaliatory tariffs. Aghion, Bergeaud, Lequien, and Melitz find that tariff exposure reduced firm employment growth and weakened firm dynamism. Bown, Conconi, Erbahar, and Trimarchi find that antidumping protection against China reduced downstream employment, wages, sales, and investment, with no significant protected-sector employment gains. Javorcik and coauthors similarly find that input-tariff and retaliatory-tariff exposure reduced online job postings.

The best reading of the employment literature, then, is not that tariffs simply “create jobs” or “destroy jobs.” Rather, tariffs redistribute employment. They may preserve jobs in directly protected firms or industries, but they also raise costs for downstream producers, invite retaliation against exporters, and reduce activity in industries that rely on imported inputs. When studies focus narrowly on protected sectors, the employment effect is more likely to appear positive. When studies account for supply chains, downstream users, retaliation, or broader regional exposure, the effect is more often negative. For instance, tariffs designed to protect the jobs of steel workers may offer some protective effect for those workers, but for every steel worker there are 80 workers in industries that use steel as an input good, and these workers would likely suffer losses due to higher input costs.

Finally, the output, productivity, and welfare literature is overwhelmingly unfavorable to tariffs. In the studies reviewed here, 25 find negative effects of tariffs on output, productivity, or welfare, while none find positive effects. This is perhaps the strongest result after prices. Across countries including Chile, India, Indonesia, Hungary, Brazil, Canada, China, and the United States, the same pattern appears: lower tariffs tend to raise productivity, while higher tariffs tend to reduce productivity or economic welfare.

The mechanism varies by context. Pavcnik finds that Chilean trade liberalization increased productivity in import-competing sectors. Amiti and Konings find that input-tariff reductions in Indonesia generated large productivity gains, especially for firms importing intermediate goods. Topalova and Khandelwal find that both output- and input-tariff reductions raised productivity among Indian firms, with input tariffs having especially large effects. Halpern, Koren, and Szeidl find similar productivity gains from access to cheaper foreign inputs in Hungary. Other studies show that tariff reductions increase product scope, induce technology adoption, improve resource allocation, or allow more productive firms to expand. The recent protectionist episodes tell a similar story from the opposite direction: higher tariffs raise costs, reduce real income, and lower welfare.

Taken together, the empirical literature tells a consistent but nuanced story. Tariffs reliably raise prices and reduce targeted trade flows. Their employment effects are more ambiguous because they depend heavily on the level of analysis: protected industries may gain, but downstream firms, exporters, and consumers often lose. The productivity and output evidence, however, points strongly in one direction. Tariffs shelter some firms from competition and may preserve some jobs in the short run, but they do so at the cost of higher prices, reduced trade, weaker productivity growth, and lower overall economic welfare.

Between January 2025 and January 2026, the number of individuals receiving SNAP benefits declined by nearly 4.3 million. Roughly 3.5 million of that decline occurred after the passage of the One Big Beautiful Bill Act (OBBBA) in July 2025.

Even after the recent drop in enrollment, SNAP participation remains approximately 1.7 million individuals above pre-pandemic levels. Meanwhile, the average cost per household (adjusting for inflation) is still about 18 percent higher than before the pandemic.

How SNAP Became Larger and More Expensive

Prior to the pandemic, the average SNAP benefit was roughly $235 per household. That changed dramatically beginning in March 2020, when Congress authorized states to provide all SNAP households with the maximum benefit regardless of income.

The expansion continued under the American Rescue Plan in 2021, which implemented a 15 percent across-the-board increase to maximum SNAP benefits.

Although many of these changes were advertised as temporary emergency measures, the Biden administration later made a substantial portion of the expansion permanent. In 2021, the USDA reevaluated the “Thrifty Food Plan” and permanently increased SNAP benefit levels by 21 percent above the normal annual inflation adjustment.

This administrative change was historically unprecedented. Traditionally, reevaluations of the Thrifty Food Plan were expected to remain cost neutral. Instead, the Biden USDA used the process to permanently ratchet benefit levels higher.

By January 2026, average SNAP benefits per household were approximately 48 percent higher than pre-pandemic levels. During that same period, the overall cost of living increased by roughly 26 percent.

Benefits therefore rose substantially faster than inflation. At the same time, participation surged. SNAP enrollment peaked at roughly 43 million individuals in 2020 and remained elevated long after the economy reopened.

Beginning in April 2020, work and work-training requirements were largely suspended. States were granted broad “flexibility” regarding eligibility verification and recertification. Interview waivers, automatic benefit extensions, and relaxed administrative oversight made it substantially easier to remain enrolled in the program.

In 2020 alone, SNAP participation increased by more than 6 million individuals.

The SNAP program also faces a chronic problem that predates the pandemic: improper payments. In recent years, SNAP erroneous payment rates have hovered around 9 to 10 percent, nearly double the rate seen seven years earlier. These erroneous payments cost taxpayers roughly $10 to $12 billion annually.

Importantly, most improper payments are not necessarily intentional fraud. Many result from administrative errors, incorrect eligibility determinations, or failures to update household information. But regardless of the cause, the fiscal consequences are enormous.

A program that in some years costs $120 billion annually cannot sustain persistent error rates near double digits without meaningful reform.

What the OBBBA Changed

The OBBBA represented the most significant SNAP reform in years. One important provision restored the traditional expectation that future reevaluations of the Thrifty Food Plan remain cost neutral. This prevents future administrations from unilaterally expanding SNAP benefit levels above inflation through administrative reinterpretation alone.

However, the legislation did not reverse the Biden-era benefit increase already implemented in FY2022.

More consequential reforms involve state financial responsibility. Historically, SNAP benefit costs were fully federally funded, while administrative costs were split evenly between states and the federal government. Beginning in 2027, states will be responsible for 75 percent of administrative costs, with the federal government covering only 25 percent.

More importantly, for the first time in the program’s history, states will also share direct responsibility for benefit costs themselves. Under the new structure, states will cover between 5 and 15 percent of benefit costs depending on their payment error rates. States with high erroneous payment rates will pay more, while states with lower error rates will pay less.

Under the previous structure, states faced weak incentives to aggressively reduce payment errors because the federal government absorbed nearly the entire cost of mistakes. The new system introduces real fiscal consequences for poor administration.

The OBBBA also tightened work requirements for able-bodied adults without dependents and reduced states’ ability to issue broad waivers. At the same time, the age threshold for work requirements was increased from 54 to 64.

Finally, eligibility rules were tightened for certain non-citizen groups, including refugees and asylees.

Why Is SNAP Enrollment Falling?

The OBBBA is clearly one important explanation for the recent decline in enrollment. Between December 2024 and July 2025, before the legislation passed, SNAP participation had already fallen by about 1 million individuals. But after enactment, enrollment declined by another 3.5 million.

The timing strongly suggests the reforms accelerated the downward trend, but policy changes are probably not the entire explanation.

As AEI scholar Angela Rachidi recently observed, some of the largest declines occurred in states like Florida and Georgia that did not previously maintain broad work requirement waivers for able-bodied adults. That suggests broader economic and post-pandemic normalization effects are also playing a major role.

Indeed, after the Great Financial Crisis, SNAP enrollment eventually declined by nearly 11 million individuals as labor markets recovered. Following the COVID-19 recovery, however, participation remained unusually elevated for years despite a strong labor market.

In that sense, some correction was inevitable. The pandemic-era welfare expansion significantly increased both the generosity and accessibility of SNAP. What we are seeing now is not simply “cuts,” but rather a gradual return toward something closer to historical norms.

Further Reforms Are Still Needed

The recent decline in SNAP participation is encouraging, but the program still remains substantially larger and more expensive than before the pandemic. Additional reforms could further improve both program integrity and long-term sustainability.

First, current reporting rules only require disclosure of payment errors exceeding $57 per individual. That threshold should be eliminated entirely so that all payment errors are disclosed and tracked. Since states will now bear part of the financial burden of erroneous payments, greater transparency would strengthen incentives to identify and prevent errors.

Second, SNAP EBT cards remain technologically outdated. Unlike modern credit and debit cards, many EBT cards still lack chip security features, making them more vulnerable to fraud and theft. Some states, including Oklahoma, have already moved toward chip-enabled EBT cards that better match modern banking standards. USDA should require all states to adopt this model.

Finally, Congress should consider aligning SNAP’s definition of “elderly” with the Social Security Administration’s full retirement age of 67. Current eligibility rules use a lower threshold that no longer reflects modern workforce participation patterns or retirement norms.

The SNAP program exists to provide temporary nutritional assistance for vulnerable households. But over the past several years, temporary emergency expansions evolved into a far more permanent enlargement of the welfare state.

The recent enrollment decline suggests that process is finally beginning to reverse. The question now is whether policymakers will finish the job.

This would be all well and good were it not for the hugely positive impact that data centers have on the economy. Ever since the digital revolution, the internet has relied on servers stored in data centers so that it can function. This means that if we were to limit their construction due to these negative side effects, real as they are, there would also be negative effects faced due to the lack of progress being made through computation. In economics, this is known as opportunity cost. And while many would argue that the opportunity cost of not building a data center is not worth it in the long run, I argue something else entirely. We can expand effective data center capacity without relying on building new ones.

In simple terms, data centers house large computers that store, process, and transmit information just like any other computer might, just with a far larger amount of memory and processing power than your typical desktop. Other than their aggregated capacity, there isn’t anything particularly special about these computers; they still have storage, RAM, CPUs, GPUs, and all of the other important parts that constitute a personal computer. This means that we could feasibly build a data center by using computers that we would use on a daily basis. So why don’t we?

It turns out that, since it is essentially impossible for a computer to run at maximum computing capacity all the time, excess capacity can be offloaded to a sort of decentralized data center. By taking and pooling this excess computing power, we can construct a network made up of thousands of personal computers. This is known as volunteer computing, and it can be used to take advantage of the positive network effects that arise when you decide to chain together a set of computers. This is already being used in various industries, such as 3D rendering with the SheepIt render farm and academia with Berkeley’s BOINC system. The most famous of these systems is known as Folding@home, which is a protein folding volunteer computing system, and in 2020, it became the largest computing system ever created, being the first system to achieve exaflop status.

There are a few ways to incentivize someone to enroll in a system like this. As with the above examples, a user could receive the benefit of having direct access to the system and its increased computing power. This, however, is a bit more of a niche incentive, only applying to a small number of people. Another simple benefit would be to pay people for the amount of computing power that they donate to the system. This is already a thing with smaller systems like the Akash and Golem Networks that pay people for providing them with compute, but due to their small client base, these aren’t yet known to be amazingly effective when it comes to making consumers extra income.

It seems strange, then, that if there are so many benefits to using these systems, they aren’t more popular. A lot of this has to do with the fact that decentralized systems also tend to lack standardization and consistency. A computer that is deploying to a volunteer computing service can only offer its computing power when the user allows it to, and when it “donates” compute, it offers an entirely unique set of specifications that the system must then find a way to mold so that it can fit into the current system.

A simple analogy would be a railroad network that is built entirely from donated pieces of wood and metal. These have not been vetted beforehand, and thus need to be reformatted and reshaped before they can be properly used. If they do not undergo this reshaping process, then there is no guarantee that they will allow the train to run on it properly. At this point, why would the railroad company want the donated material when they could harvest it themselves? The reformatting process, the costs associated with not knowing when people will donate, and the general uncertainty involved in the system make it almost not worth it.

But certainly, the railroad companies would be unwise to reject all donations. Not everything offered is junk. Those pieces that need only to be cleaned of their rust, sanded down, have lower costs associated with them than other, more mangled bits. This is true as well of certain computers. Not all are created equal, not all are used equally, and as a result, using the system is costly. For example, it is considered standard practice in a volunteer computing system to perform every calculation twice to verify that the first result was not mistaken. This, among other problems, already hinders efficiency quite a bit, thus reducing the amount of incentives the systems would be able to give to users to enroll.

The best way to reconcile this would be to have a set of standards for those participating in the system. This would look like minimum hardware requirements, uptime guarantees, and other software-side systems that could more properly handle these constraints. Then, by interfacing with a centralized network, these volunteer computing systems would be able to perform those highly parallelizable computations that larger systems should not be wasting their time and capacity on, thus freeing data centers from having such high requirements and offloading some of their benefits to consumers.

It is important to note, however, that this would not serve to replace data centers, but instead to bolster them. Not all tasks qualify to be solved with volunteer computing; some need a greater degree of centralization, either due to process constraints or limitations due to the size of the data set. In these cases, the scope of the data further disqualifies it from participation due only to the massive cost associated with transferring the data, making local data centers necessary. This volunteer computing solution would only be able to solve those simple tasks that data centers should not be wasting their time on.

From an economic perspective, the benefits are twofold:

1. Data centers’ capacity expands, while diminishing the increase in local negative externalities associated with them.

2. Consumers are paid for their volunteer computing power, meaning that the social cost is further offset by the benefits faced by the consumer.

The United States is on an unsustainable fiscal trajectory. Federal debt held by the public has risen sharply over the past two decades and is projected to continue climbing indefinitely under current policy. At the same time, rising interest rates have begun to materially increase the cost of servicing that debt. Interest payments are now among the fastest-growing components of the federal budget, steadily crowding out other priorities. This dynamic is not merely a long-run concern but is already reshaping the composition of federal spending today.

Looking ahead, the federal government faces a looming fiscal crisis. As Veronique de Rugy has emphasized, the eventual depletion of the Social Security and Medicare trust funds will force policymakers to confront large and immediate funding gaps. When those trust funds run dry, benefits can only be paid out of incoming revenues, requiring abrupt benefit adjustments, significant new borrowing or large-scale fiscal consolidation. Financial markets, anticipating these pressures, may demand higher interest rates on U.S. debt, compounding the problem and accelerating the need for policy action.

In such an environment, federal policymakers will be forced to make difficult choices about how to close the gap between spending and revenues. But the range of politically feasible options is likely to be constrained. Major entitlement programs such as Social Security and Medicare are widely viewed as the “third rail” of American politics, making substantial reforms politically unpopular. Similarly, reductions in defense spending are often limited by national security considerations. While tax increases are frequently discussed, tax-based consolidation alone will not meaningfully improve our fiscal trajectory and could worsen conditions by hampering economic growth.

This leaves other areas of the federal budget as more plausible targets for adjustment. Among these, federal aid to state and local governments stands out as a particularly significant, and potentially vulnerable, category. According to historical budget data from the Office of Management and Budget (OMB) federal grants to states currently exceed $1.2 trillion annually and have grown substantially as a share of total federal spending. In the years following World War II, such grants accounted for just 2% to 5% of federal outlays. Today, they represent roughly 17% to 19%, reflecting a long-term shift toward greater federal involvement in areas that were traditionally state and local functions.

As the Hoover Institution’s John Cogan observes, the scope of federal spending has expanded to the point where “no state or local activity is beyond the reach of the federal government’s check-writing machine.” Federal funds now support not only major entitlement programs administered at the state level, but also a wide array of activities once considered firmly within state and local jurisdiction, from infrastructure and education to recreational and community development projects. This breadth underscores the extent to which state budgets have become intertwined with federal fiscal policy.

The implication is straightforward but often underappreciated: If and when the federal government is forced to retrench, states will not be insulated from the consequences. On the contrary, they are likely to be directly affected. With roughly 35-36% of state revenues coming from federal sources, any significant reduction in federal aid would create substantial budgetary pressures for state governments. These pressures could manifest as spending cuts, tax increases or both, potentially during periods of broader economic stress.

Given these risks, the question is not whether states will be affected by federal fiscal consolidation, but how well prepared they are to absorb it. Some states have built stronger fiscal foundations, maintaining lower spending levels, better-funded pension systems and larger budget reserves. Others remain more exposed, with higher spending commitments and greater reliance on federal transfers. Yet despite these differences, there is currently no widely used, comprehensive framework for evaluating state-level fiscal resilience to a reduction in federal support.

This gap motivated me to create a Fiscal Resilience Ranking. By systematically assessing key dimensions of state fiscal health, including spending intensity, liability funding, fiscal space, rainy-day reserves, revenue volatility and dependence on federal aid, this index provides a picture of how prepared each state is for a future in which federal support may be less certain.[1]

This is a first attempt at building a fiscal resilience ranking and may not be a definitive or final measure. There are undoubtedly dimensions of fiscal health not fully captured here, and reasonable people may disagree on how these variables should be defined or weighed. I welcome feedback and suggestions for ways that I can refine this framework.

The map below shows all 50 states with a fiscal resilience score. The lowest-scoring state is New Mexico with a score of 3.91 out of 10, and the highest-scoring state is Texas with a score of 7.53 out of 10.

The table below ranks the 50 states from 1 to 50 based on their respective scores, along with the six sub-scores that make up the overall ranking score.

The overall ranking scores are the average score over the six sub-scores. The six sub-scores are defined as follows:

Spending Intensity
This measure captures state government spending per capita, reflecting the overall size and cost of government relative to its population. States with higher spending levels may face greater difficulty adjusting budgets in response to fiscal shocks, thereby reducing their resilience.

Funded Liabilities
This metric measures the share of long-term obligations, such as pensions and other promised benefits, that are backed by existing assets. Higher funded ratios indicate stronger fiscal health and a reduced need for future budgetary tradeoffs or abrupt adjustments.

Fiscal Space
Fiscal space is defined as the combined state and local tax burden as a share of state income, capturing how much additional revenue capacity a state has. States with already high tax burdens may be less able to raise additional revenue without risking economic distortion or diminishing returns.

Rainy-Day Fund
This measure estimates how many days a state could continue funding operations using only its budget stabilization reserves. Larger reserves provide a buffer against revenue shortfalls or external shocks, thus enhancing short-term fiscal resilience.

Federal Dependency Rate
This metric calculates the share of state revenues derived from federal transfers, grants and aid. States with higher dependence on federal funding are more vulnerable to federal fiscal retrenchment and policy changes.

Revenue Volatility
Revenue volatility is the standard deviation of year-over-year growth in total state tax collections. This metric captures the sensitivity of state revenues to economic conditions. States with more volatile revenue structures are more prone to cyclical revenue swings, increasing the likelihood of fiscal stress during downturns and reducing overall fiscal resilience.

[1] Pension funding metrics are based on reported actuarial values, which rely on state-specific discount rate assumptions and may understate the true economic value of liabilities.

The Yale study cited here was published in 2020 in The Lancet. If we overlook the questionably biased section headings throughout the paper, such as “The life-saving potential of Medicare for All,” we can start to decipher where the 68,000 figure comes from.

The authors claim that uninsured individuals experience a 40% elevation in age-specific mortality risk. Using this estimate, they calculate that universal coverage would save 68,531 American lives on an annual basis. They also note that these lives are mostly relatively young people, as individuals older than 64 are already covered by Medicare.

So where did the authors get the 40% figure? It comes from a 2009 study based on a survey of 9,004 individuals, 2,350 of whom were uninsured. Over the 1988-2000 observation period, uninsured individuals were found to be 40% more likely to die than the insured survey participants.

Contrast this finding with that of a larger survey taken over a slightly longer period. This 2009 study published in Health Service Research surveyed 672,526 individuals between 1986 and 2002. The author found that “the risk of subsequent mortality is no different for uninsured respondents than for those covered by employer-sponsored group insurance at baseline (hazard ratio 1.03, 95 percent confidence interval [CI], 0.95–1.12).”

Notice that the confidence interval crosses below 1, suggesting no statistically significant relationship between insurance coverage and mortality. The author concludes:

The Institute of Medicine’s estimate that lack of insurance leads to 18,000 excess deaths each year is almost certainly incorrect. It is not possible to draw firm causal inferences from the results of observational analyses. … The answer suggested by the evidence presented here is that there would not be much change in the number of deaths in the United States as a result of universal coverage.

Moving on from survey-based studies, a 2014 nonrandomized study comparing health outcomes in Massachusetts (which had near-universal health coverage) with control-group counties found that insurance coverage reduced all-cause mortality by 2.9 percentage points. The authors were careful to note, however, that the study design was subject to unmeasured confounding variables.

Even if we take that 2.9 percentage points at face value, the reduction would imply about 2,189 fewer deaths with full-population health coverage, not quite the 68,531 annual deaths claimed in the Yale study.

Observational survey data and nonrandomized studies are not well-suited for measuring the complex relationship between insurance coverage and mortality, which is subject to confounding variables and other factors. Perhaps the best methodological approach is to review randomized control trials (RCTs).

One such RCT was published in the Quarterly Journal of Economics in 2021. Around 2017, the Internal Revenue Service (IRS) started a pilot program in which it sent letters to uninsured taxpayers under the individual mandate of the Affordable Care Act. Because the pilot led to increased coverage, this group of taxpayers was used to explore the causal relationship between health insurance and mortality.

Two years after receiving the IRS letters, the rate of mortality among 45- to 64-year-olds declined by 0.06 percentage points compared to the control group, or one fewer death for every 1,648 individuals. Interestingly, the RCT’s authors found no evidence of changes in mortality for younger individuals. This finding is noteworthy because the Yale study said that excess deaths were “predominantly the lives of young people.”

Another RCT study published in 2013 observed the Oregon state Medicaid lottery that allocated a limited number of Medicaid slots to poor, able-bodied, uninsured adults age 19 to 64. The lottery winners showed a lower death rate of 0.8% compared to lottery losers, representing a “dose-adjusted” reduction of 0.13 percentage points. But the finding was not statistically significant.

The claim that tens of thousands of Americans die annually due to lack of insurance rests on fragile empirical ground. Estimates like the Yale study’s 68,000 figure hinge on small observational studies that struggle to isolate causality, while larger samples and randomized evidence point to far smaller, and often statistically insignificant, effects.

None of this implies insurance is unimportant, but it does mean the magnitude of its effect on mortality is frequently overstated. Policymaking should be guided by careful evidence, not attention-grabbing figures that exceed what the data can credibly support.

A new paper by Finbar Curtin and Matthew Burgess, “The Empirically Inscrutable Climate-Economy Relationship,” argues that this dispersion is not a temporary problem awaiting better data or cleverer econometrics. It is, instead, a fundamental and irreducible feature of the enterprise. Their conclusion is uncomfortable: we cannot reliably estimate the macroeconomic damage from climate change using historical data.

The Identification Problem at the Core

Most empirical climate-economy models follow a similar structure. Researchers estimate how temperature (or other climate variables) has historically affected GDP, then feed projected future warming into that relationship to generate estimates of future economic damages.

The problem, Curtin and Burgess argue, is that this relationship is not stable or uniform, but varies dramatically across both space and time.

Countries with the same average temperature can have entirely different economic responses to climate variation. El Salvador and Iraq, for example, may share similar climates but have radically different economic structures, institutions, and adaptive capacities. Pooling them into a single regression implicitly assumes a common “damage function” that simply doesn’t exist (see Figure 3 below).

Time adds another layer of complexity. As countries grow richer, they adapt. Air conditioning, infrastructure, migration, and sectoral shifts all change how temperature affects economic output. That means the relationship between climate and GDP is itself evolving, precisely the kind of instability that undermines standard econometric identification.

Economists typically deal with this kind of heterogeneity by imposing simplifying assumptions. But here’s the catch: if you assume away too much variation, your model becomes misidentified. If you try to fully account for it, you run out of degrees of freedom and can’t estimate anything meaningful.

There’s no econometric trick that resolves this tradeoff. That’s why the authors describe the relationship as “empirically inscrutable.”

When a Few Data Points Drive the World

If the identification problem weren’t enough, the paper highlights another issue that should make anyone uneasy: extreme sensitivity to a handful of observations.

In one prominent study (Burke et al. 2015), removing just six data points out of more than 6,000 reduces the estimated climate effect by about 25 percent.

Even more striking is the example of a now-retracted Nature paper, where erroneous data from a single country, Uzbekistan, accounted for roughly two-thirds of the estimated global climate damage effect (see the 3 figures below). Mistakes do, of course, happen, but the deeper question is: how can one small country plausibly drive global conclusions in the first place?

The answer lies in how pooled regressions work. When countries are combined into a single statistical model, idiosyncratic events, like Rwanda’s genocide or Armenia’s post-Soviet collapse, can end up disproportionately influencing the estimated “global” relationship between climate and growth.

In effect, these models can imply that how a small, crisis-stricken country responds to a hot year tells us something meaningful about how the United States or Germany will respond to long-run climate change. That’s a leap few would defend if stated plainly.

So What Should We Do?

It’s important to be clear about what Curtin and Burgess are not saying. They are not claiming that climate change is harmless or that reducing emissions lacks value. The existence of a negative carbon externality remains well-established.

What they are challenging is the confidence with which economists, and especially policymakers, treat specific numerical estimates of climate damage.

If the underlying relationship cannot be reliably identified, then there is no single “correct” social cost of carbon. The wide range of estimates is not a temporary inconvenience but reflects a deep uncertainty that cannot be eliminated with better data or more sophisticated models.

Rather than pretending that we can fine-tune climate policy based on precise damage estimates, we should acknowledge the limits of our knowledge. This pushes us toward a framework of decision-making under deep uncertainty, where robustness, resilience, and flexibility take precedence over point estimates and optimal control.

There’s a broader takeaway here, one that extends beyond climate economics.

Policymakers often demand precise numbers, whether it’s the fiscal multiplier, the elasticity of taxable income, or the social cost of carbon. Economists, in turn, are tempted to provide them, even when the underlying uncertainty is substantial.

Only last year the Congressional Budget Office CBO published a paper that does precisely this, concluding that climate change would cause GDP to be 4 percent lower in 2100 than if temperatures had stayed constant. The CBO takes these studies into account when producing its long-term forecasts.

But precision is not the same as accuracy, and when models are fundamentally fragile, presenting point estimates can be more misleading than helpful. The Curtin-Burgess paper is a reminder that some questions may not yield to empirical methods as cleanly as we would like. In those cases, humility is not a weakness, it’s a necessity.

Trillion-dollar decisions deserve more than false precision.