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.

Addressing the concerns

Unfortunately for proponents, the arguments against the construction of new data centers are rather strong. These massive, windowless blocks of concrete come with a myriad negative externalities, all of which directly impact the local community. From increasing the cost of living to providing measurable health risks, data centers do seem to have a lot of ground to cover before they can find their way back into the good graces of the general public.

Equally unfortunate, however, is the use of misinformation to alter public opinion. For example, one of the most common points of evidence used by those who oppose the construction of data centers is that they use a great deal of water, and that doing so could have extraordinarily harmful effects on the environment. The issue is that many people are continuing a discourse that started with Karen Hao’s Empire of AI, in which she made the claim that a single data center used 1000 times more water than a single city. The real number, of course, is far lower, sitting around 0.22 times, but now that their water usage has entered the conversation, it is difficult to move forward without it.

Other common criticisms of data centers, such as that they increase the cost of land and electricity use, are largely true, but should not come as a surprise to a careful, economically minded thinker. Data centers are not unique in that they increase demand for the resources that they consume, and when these resources are also consumed by households, households will, as a result, experience the effects of that rise in demand through higher prices. Once again, Virginia, especially Northern Virginia, has been at the forefront of the coming change. In fact, it is one of the only places in the United States that has shown dramatic and definitive evidence of data centers affecting rising energy costs. Additionally, residents of Loudon County, Virginia, have begun to complain about a low-frequency hum that is constantly being emitted by the data centers, which people now worry may cause health problems.

The list goes on, and if we are to consider it fully, it is challenging not to agree with the data center opponents, who now seek to curb their development. The story not being told, however, is that most of these objections apply similarly to other productive sectors. Large buildings, factories, and office complexes all use a large amount of electricity and water, pollute as a result of their output, and raise land prices. There is a reason that large cities tend to be costlier to live in than rural areas. Scarcity is the law of the land, not some mystical force created by a confluence of computer chips. Thus, it seems that the objection to data centers does not merely come from what they consume throughout their operation, but also the product itself.

Data centers produce more than just AI.

At least, this should be the case. People tend to correlate data centers with AI language models, such as ChatGPT and Claude. With this in mind, it would stand to reason that people would then have similar feelings regarding AI and the data centers that maintain it. The issue is that this isn’t the case. A poll found that 48% of people viewed AI favorably, while 46% viewed it unfavorably. This, when compared to the 35% in favor of new, local data center construction, is a clear mismatch of preferences, where people enjoy the product yet do not appreciate the effort necessary to create it.

But data centers do not only produce AI. It is no exaggeration to say that data centers have become the very backbone of society, and that without them the entire world would collapse. As the centerpiece of the information sector, data centers are the reason you are able to use Google Drive, invest online, order food with DoorDash, and so many more things that have made our lives more convenient. On top of this, 46% of the world’s data centers are in the United States, and a recent report from S&P Global shows that they contributed as much as half a point to our GDP growth in the second quarter of 2025.

Thus, it is clear, at the very least, that data centers do provide a great amount of good to society. How, then, should we address the issues that they present to society?

We should be focusing on the tax side, instead.

From an economic perspective, data centers do pose a problem. More than almost any other industry, they receive a large amount of tax breaks that allow them to more easily develop. The tax breaks vary by state, but among them in Virginia is a sales tax exemption on the purchase of qualifying equipment and software, reducing costs by nearly $1 billion in 2023, and local property tax exemptions of up to 89%. This might sound good, perhaps even innovative, but the issue is that these tax breaks are actually distortionary, essentially granting data centers an upper hand that they otherwise wouldn’t have had. This means that on top of the already existing negative externalities that they produce, the government is also essentially subsidizing this social cost and enhancing the negative effects.

The solution, then, is simple. We should not ban the construction of data centers, nor should we disproportionately tax them to discourage their construction. Instead, we should remove these tax breaks and have data centers step back into the competitive market so that they can reinforce our infrastructure without the distortions of government intervention.

The Governor of Maine recently signed into law a budget that includes a 2 percent surcharge on income over $1 million. This comes just weeks after Washington state imposed a 9.9 percent tax on income above $1 million. Other states such as Minnesota and Maryland have imposed additional capital gains surtaxes on high earners in recent years.

Many blue states now have top marginal income tax rates (combined federal, state, local, FICA, and NIIT) that are above 50 percent, and this list looks set to grow further.

To test whether additional state taxes on high earners notably increase state tax revenues, we can observe the revenue trends of two states that have imposed millionaire taxes in recent years: New York and Massachusetts.

In 2021, then-New York governor Andrew Cuomo, raised the state’s top marginal tax rate on earners making over $1 million to 9.65 percent from 8.82 percent. He also introduced two new brackets for those earning over $5 million and $25 million, with a top rate of 10.9 percent. In New York City, taxpayers also have to pay the 3.9 percent city income tax, for a top combined state and city rate of 14.8 percent.

Two years later in 2023 Massachusetts imposed a new 4 % surtax on incomes over $1 million added to its existing 5 percent flat tax (for a top rate of 9 percent). Voters approved this measure via referendum in 2022, and politicians promised a surge of new revenues.

To remove the distortion of rising prices, the first metric to review is real income tax revenues. Using a CPI deflator, the two bar charts below show changes in revenues over time in 2018 dollars. The red dashed line shows the level of revenue in the year before the new tax was implemented. Green bars are real revenues before the tax change and orange bars show real revenues after the tax change.

In New York, after raising tax rates on high earners in 2021 real revenues seem to spike in 2022 before falling below pre-tax reform levels in the 2023-25 period. In contrast, Massachusetts saw an immediate drop in real revenues the year the tax first applied (2023) before slowly recovering to pre-reform levels by 2025. Neither trend demonstrates a notable upswing in income tax revenues that we might expect to see from the windfall policymakers often promise.

An alternative way of measuring changes in income tax revenues over this period is to observe how revenues as a share of state gross domestic product change (GDP) over time. The first line chart below shows this for New York. The orange dashed line shows the average income tax revenue to GDP in the pre-reform period. Like the bar charts above, we see a spike in 2022, but then 2023-25 revenues are notably below pre-reform levels.

The grey shaded area shows the post-reform period.

The second line chart below shows the same metric for Massachusetts. In this case, we see an immediate drop in revenues in the first year that the millionaire tax applied before recovering to pre-reform levels by 2025.

To move beyond simple before-and-after comparisons, we can construct a counterfactual for each state using a synthetic control approach. The idea is straightforward: instead of asking whether revenues rose or fell in isolation, we ask how they performed relative to a “synthetic” version of the state built from similar states in the region that did not adopt surtaxes. In this case, Connecticut and Maine serve as donor states, with weights chosen to minimize pre-reform differences in income tax revenue as a share of GDP.

When these weights are optimized to best match pre-treatment trends, the synthetic controls track both New York and Massachusetts reasonably closely prior to their respective policy changes. This gives some confidence that the post-reform divergence is informative. After New York’s 2021 tax increase, its revenue-to-GDP ratio falls persistently below its synthetic counterpart. Averaging over 2021–2025, New York collects about 0.33 percentage points of GDP less in income tax revenue than its counterfactual benchmark would predict.

Massachusetts shows a similar, though more muted, pattern. Following the 2023 surtax, revenues initially drop below the synthetic control and only partially recover. Over 2023–2025, Massachusetts collects roughly 0.14 percentage points of GDP less than its synthetic benchmark.

In both cases, the direction of the gap runs counter to the “windfall” narrative often used to justify these policies. While this is a simplified exercise, using a limited donor pool and not a full formal synthetic control, it reinforces the broader pattern: higher marginal tax rates on top earners have not translated into sustained gains in revenue relative to underlying economic activity.

The best recent quasi-experimental evidence from California’s 2012 tax hike shows just how quickly high earners adjust. Rauh and Shyu (2024) find that nearly half of the projected revenue gain disappeared within a year, and over 60 percent within two years, driven primarily by behavioral responses among those who stayed, not just by out-migration. Estimated elasticities of 2.5–3.2 imply that high-income taxpayers are extremely responsive to marginal tax rates, making static revenue estimates deeply misleading.

Once we account for the full set of avoidance and adjustment margins: income shifting across tax bases, changes in organizational form, and intertemporal timing, the tax base becomes far more elastic than naive models assume. The result is a much flatter Laffer curve: tax rate changes have only modest effects on revenue. In the most realistic calibrations, the revenue-maximizing top rate is close to 40 percent (combined). Notably, several high-tax states already impose combined federal, state, and local top marginal rates at or above this level, exceeding 50 percent in some cases, suggesting that further rate increases are especially unlikely to generate meaningful additional revenue.

One notable behavioral response that state policymakers often overlook is that state residents can vote with their feet and take their income with them. The charts below show the net migration of high earners in both New York and Massachusetts during the 2018-2023 period. High earners in this case are defined as individuals making more than $200,000 a year. New York lost, on net, 176,000 high earners in 5 years who took $50 billion of income with them. Massachusetts lost 33,000 high earners and $9.3 billion in income.

When a small number of high-income taxpayers account for a disproportionate share of the tax base, even modest net outflows can materially erode expected revenues. This helps explain why the revenue effects of higher top marginal tax rates often fall short of projections.

The evidence from New York and Massachusetts provides no support for the claim that higher marginal tax rates on top earners generate large and sustained revenue gains. In both states, the initial promise was straightforward: raise rates on a narrow group of high-income taxpayers and use the proceeds to finance new spending. Yet the data tell a more complicated story. Real revenues are volatile and, in both cases, fail to show a clear and lasting increase after the policy change. When measured relative to state economic output, the picture is even less favorable, with revenues either falling below prior levels or underperforming reasonable counterfactual benchmarks.

Imposing additional surtaxes on high earners may be politically appealing, but the data and evidence suggests that this tax base is too mobile and too responsive to deliver the revenue windfalls that policymakers often promise. If these policies do not reliably generate additional revenue, it raises an obvious question about their primary objective.

After stabilizing briefly in December, public debt accelerated upwards in Q2, crossing the $31 trillion point in February, as figure 1 illustrates. As a share of GDP, the public debt ratio is 100%. This is the second-highest level on record (the highest level was in 1946 at 106%), while the Congressional Budget Office forecasts our debt trajectory will reach a record high within the next three years.

At this level of public debt, the U.S. economy forgoes about 0.8 percentage points in potential economic growth, based on a synthesis of nearly 200 estimates in the empirical literature. In other words, real economic growth in 2026 would be about 3.2% rather than the forecasted 2.4% if our debt ratio were stable at 75% of GDP, as it was in 2019.

Halfway through the fiscal year, the federal government has spent a total of $3.65 trillion, while bringing in a total of $2.43 trillion in receipts, according to the latest Monthly Treasury Statement. Federal revenues are slightly higher than last year, but spending is also higher than in prior years. As a result, the budget deficit for the first six months of FY 2026 is already $1.17 trillion. As figure 2 shows, this six-month budget deficit is smaller than last year’s, but still notably larger than in prior years.

Interest Rates and Interest Payments

One major federal outlay that has played a key role in driving spending levels higher in recent years is the growing burden of federal interest payments on the debt. Over the past 12 months, the average interest rate on U.S. government debt as a share of GDP has hovered around 3.3%. This rate is significantly higher than the 2000-2021 average of 1.76%, as shown in figure 3. As both the public debt ratio and the interest rate on U.S. Treasurys move higher, interest expenses crowd out a larger share of the federal budget, leaving less revenue to be distributed to other policy goals and programs. The larger budget deficit also crowds out economic growth by reducing private savings and investment.

As Treasury yields remain stubbornly high, interest payments continue their upward trajectory, resulting in a larger share of federal revenues being spent on servicing the debt. Figure 4 shows that, year to date, almost 21 cents of every dollar of tax revenues collected was spent on servicing the debt. This proportion is similar to the prior two years, but notably larger than in 2023 and more than double the share of revenues spent on interest in 2022.

Maturity Structure of Public Debt

The maturity structure of public debt is a critical indicator of fiscal risk, revealing how soon the government must refinance its obligations and how vulnerable it is to changes in interest rates. The U.S. debt maturity structure is relatively short-term, meaning most of the debt is regularly rolled over into short-to-medium-term Treasuries. Figure 5 shows the average maturity of public debt over time since 2000. Historically, the average maturity of U.S. public debt is 64 months or roughly 5 years, fluctuating between a low of 49 months and a high of 75 months. Looking at the most recent quarterly data (Jan-Mar), the average debt maturity is 70 months, or just under 6 years—unchanged from the last quarter.

Looking at the distribution of public debt by different maturities, we can see that 33-34% of all public debt matures within 1 year or less, while another 34-35% matures within 1 to 5 years. Figure 6 reveals that the debt maturity distribution has changed very little over time, with about 70% of debt consistently maturing within 5 years. The share of public debt with a maturity of more than 10 years has increased slightly, while the share with maturity between 5 and 10 years has decreased slightly over time.

Using the latest data from the Monthly Statement of the Public Debt (MSPD), we can see that in the coming quarter (Apr-Jun), the Treasury will roll over a little more than $5.76 trillion in maturing Treasurys, or roughly 18% of all public debt. The fact that almost one-fifth of the entire debt stock gets rolled over in a single quarter demonstrates just how short-term the maturity structure of U.S. public debt is. Rapid and persistent increases in interest rates can have serious adverse consequences for federal budget sustainability—a risk that becomes more significant as the debt stock continues to grow.

Who Holds Public Debt?

Understanding who holds the U.S. public debt reveals not only the financial structure behind government borrowing, but also the potential economic and geopolitical vulnerabilities tied to our fiscal position. Up through the Great Financial Crisis, the dominant purchaser of U.S. Treasurys was foreign investors (foreign governments, central banks and foreign financial institutions). Between 2008 and 2015 about half of U.S. public debt was held by these foreign investors. For the past decade, however, the share of public debt held by foreign investors has declined, with foreign investors now holding just 30% of public debt. Figure 7 shows the change in who holds U.S. public debt over the past 26 years, focusing on the four largest purchasers.

Notably, mutual funds now hold more Treasury debt than the Federal Reserve bank. However, the Federal Reserve restarted quantitative easing (QE) in December—buying $40 billion of U.S. short-term Treasury debt every month. Since resuming QE, the Federal Reserve has absorbed about a third of the value of all newly issued debt during that period.

As of the latest Treasury Bulletin data (March report), foreign investors held 30% of U.S. public debt. Other major holders include the Federal Reserve with 14%, mutual funds with 16%, and money markets with 11%. State and local governments and depository institutions each hold about 6%. The pie chart (figure 8) breaks down the holdings of U.S. public debt.

The U.S. Balance Sheet Remains Structurally Vulnerable

As FY 2026 reached its midpoint, the federal government now finds itself navigating a precarious fiscal landscape—marked by historically high debt levels, growing interest costs and a heavy reliance on short-term refinancing. While revenues have improved modestly, they remain insufficient to close the widening gap between spending and receipts. With nearly one-fifth of public debt maturing in the next quarter alone, and interest payments claiming a growing share of the federal budget, the structural vulnerabilities of the U.S. balance sheet are becoming more difficult to ignore.

To monitor the fiscal outlook in real time, I’ll be updating “The Quarterly Ledger” series every quarter. Each edition will track key changes in the federal balance sheet, including public debt levels, budget deficits, interest payments and different measures of sustainability. Follow along for ongoing analysis and data-driven insights into the nation’s evolving fiscal position. You can read last quarter’s update here.

Except that the case was never really closed.

Years ago, Chris Edwards and I looked at the same data everyone else was citing and noticed something awkward. We had this chart:

And we wrote the following:

“Supporters of the EITC often point to the strong gains in participation of single mothers in the late 1990s as evidence of the EITC’s benefits. But while the number of EITC recipients soared between 1987 and 1994, the number was flat in the late 1990s, as shown in Figure 1. Yet the years from 1994 forward were precisely the years that labor force participation by single mothers was growing strongly. That suggests other factors caused much of the participation increase in the late 1990s—perhaps the strong economy at the time and welfare reforms that increased work requirements. Note that with the weaker economy after 2000, the participation of single mothers fell substantially, suggesting that it is the economy—not the EITC—that mainly drives participation changes.”

So we were skeptical about the EITC’s supposed work benefits based on the 1990’s experience, as were some other researchers such as Lawrence Mead. But the dominant narrative was that the expensive program was some sort win for everyone in the economy.

Now comes Adam Looney at Brookings with a formal replication of the four most influential studies behind the conventional view. What makes Looney’s paper different is not just the question he asks but the variable everyone else forgot to include, and the rigor with which he proves it was missing.

The standard approach in the influential studies of the 1990s was to compare employment trends across groups of mothers defined by the number of children. Mothers with two or more children received a much larger EITC expansion in 1993 than mothers with one child. So when mothers with two or more children showed larger employment gains in the years that followed, researchers concluded the EITC caused it or caused most of it. The logic seemed airtight.

The problem is that mothers with two or more children differed from mothers with one child in another critical way that those studies did not adequately account for: they were far more likely to be on welfare before the reforms began. Looney’s data show that before welfare reform hit, roughly 26.4% of mothers with two or more children were on welfare and not working, compared to about 12.8% of mothers with one child. When welfare reform tightened access to cash assistance, imposed new administrative hurdles, and pushed caseworkers to reduce caseloads, the families most entangled with the welfare system felt the greatest pressure to find work. Those were precisely the families with more children.

Previous studies sought to address the welfare reform concern by including state-level controls, essentially asking whether states with tougher welfare policies exhibited different employment trends. Looney shows why that is insufficient. The relevant variation was not which state enacted which policies, but which families within any given state were most exposed to those changes. Welfare reform operated at the caseworker level, proportionally to how dependent a family was on the system. State-level controls cannot capture that within-state variation.

Looney’s fix is a single additional control: an interaction of year dummies with each individual’s predicted pre-reform welfare exposure, estimated entirely from data before the reforms began. When he adds this one control to the specifications used in four of the most-cited studies in the literature, the estimated employment effect of the EITC collapses to zero in every case.

But Looney does not stop there, which is what separates this paper from a simple replication exercise. He runs two additional tests to rule out the concern that his welfare exposure control is simply absorbing the true EITC effect rather than correcting for bias.

The first is a reweighing exercise. He reweighs the one-child mother group to match the welfare-exposure distribution of the two-or-more-children group. Once the two groups are compositionally balanced on pre-reform welfare exposure, their employment trajectories become nearly identical. The divergence that the canonical studies attributed to the EITC disappears entirely when the groups are made comparable on the one characteristic that the original designs ignored.

The second is a placebo test. He constructs artificial treatment and control groups from within the same EITC eligibility category, meaning groups that by construction faced identical EITC incentives and for whom the true treatment effect must be zero. That nonetheless finds large spurious effects, and the size of those spurious effects is directly proportional to the difference in pre-reform welfare exposure between the placebo groups. The actual published estimates from the canonical literature fall on exactly the same line as the placebo estimates. In other words, knowing nothing about EITC policy and only knowing the baseline welfare exposure gap between the two groups is enough to predict what the published studies found.

He also runs a simulation that randomly removes mothers from welfare, with no differential effect by the number of children or any other characteristic, at rates matching the actual caseload reductions observed in the data. That simulation, with no EITC variation whatsoever, reproduces the canonical difference-in-differences estimates almost exactly.

As a final step, he shows that the same bias afflicts an entirely separate literature on welfare time limits that uses different treatment groups and an entirely different identification strategy. That those estimates collapse in the same way when welfare exposure is controlled for reinforces the conclusion: the problem is not specific to any one study’s design. It is a systematic failure to account for differential exposure to welfare reform that ran through an entire generation of research.

This is impressive work. But the data have always pointed in this direction, and I am glad that we now have a chance to have a better policy conversation without the reflective worshipping of the EITC. The EITC was never the miracle program its advocates described.

Unfortunately, the place where I point out that EITC is also a very expensive program, one of the reasons why so many people don’t pay any income tax. It is also a program that suffers from a massive rate of overpayments and some fraud. Finally, it creates some disincentives to work too.

While the credit doesn’t meaningfully boost labor supply, the phase-out part, where benefits are withdrawn as income increases, does discourage additional work. Chris and I pointed this out too.

“To summarize, people have an incentive to reduce hours worked in both the flat and phase-out ranges of the credit. As it turns out, about three-quarters of people taking the EITC are in those two ranges where the work incentives are negative.¹⁷ So economic theory indicates that a large majority of people taking the EITC have an incentive to work less, not more. Based on this factor, a Tax Foundation simulation found that the EITC reduces overall U.S. output and employment.”

For more on this, please read our piece with Chris Edwards. It turns out it is not that outdated after all.

There is something fitting about how The Wealth of Nations ends.

After hundreds of pages on trade, markets, and institutions, Smith closes with a warning—one that feels less like a conclusion and more like a diagnosis. The same commercial society that generates wealth also enables governments to access it more easily, and with fewer constraints than before. The mechanism that makes this possible is public debt.

In this final chapter, Smith takes aim at a series of arguments used to justify government borrowing—arguments that remain familiar today.

The Political Economy of Borrowing

Smith begins with a simple but underappreciated point: governments, regardless of type, do not typically save in good times.

The parsimony which leads to accumulation has become almost as rare in republican as in monarchical governments… The taste for some sort of pageantry, for splendid buildings, at least, and other public ornaments, frequently prevails as much in the apparently sober senate-house of a little republic as in the dissipated court of the greatest king.

This lack of restraint in peace creates a predictable problem in war. In need of immediate funds, governments cannot wait for the slow collection of newly imposed taxes; the answer is to borrow.

But what starts as a mechanical, temporary expedient does not remain one. Over time, borrowing becomes the preferred strategy of government—not because it is economically superior, but because it solves a political problem.

To see why, it is necessary to look at how borrowing compares to taxation, particularly in times of war.

Borrowing, Taxation, and Fiscal Illusion

Smith’s explanation for why governments borrow rather than tax—especially during war—reveals the core of that political advantage.

They [the politicians] are unwilling for fear of offending the people, who, by so great and so sudden an increase of taxes, would soon be disgusted with the war… The facility of borrowing delivers them from the embarrassment which this fear and inability would otherwise occasion.

Further, the physical and economic costs are not borne equally throughout society. Those in the capital and remote areas are insulated from the war and may even get some enjoyment from it:

In great empires the people who live in the capital, and in the provinces remote from the scene of action, feel, many of them, scarce any inconveniency from the war; but enjoy, at their ease, the amusement of reading in the newspapers the exploits of their own fleets and armies.

The result is more frequent and longer wars than would occur if citizens bore the full fiscal cost. If wars had to be financed entirely through current taxation, the public would feel the full burden immediately:

Wars would in general be more speedily concluded, and less wantonly undertaken. The people feeling, during the continuance of the war, the complete burden of it, would soon grow weary of it...The foresight of the heavy and unavoidable burdens of war would hinder the people from wantonly calling for it when there was no real or solid interest to fight for.

This is the core of the mechanism. Borrowing allows the public to enjoy the benefits of war without bearing its full cost in the present—a dynamic later described as fiscal illusion. The result is predictable: wars become easier to sustain, more frequent, and longer in duration, with their burden shifted onto future taxpayers.

The same dynamic, of course, is not limited to war. It occurs whenever governments can expand spending on politically popular programs without tying those increases to current taxation.

Once borrowing becomes politically easier than taxation, it is not surprising that a set of arguments emerges to justify and sustain it. Smith addresses several of these directly.

Fallacy #1: “We Owe It to Ourselves”

One of the most common defenses of public debt is that it is harmless when held domestically: “it is the right hand which pays the left,” goes the defense.

Buying our own government’s debt means that domestic capital has been redirected away from more productive, market uses and toward government uses.

Yes, interest payments flow back to domestic lenders. But those lenders are not receiving some unique gain from government borrowing. They are receiving returns that, absent the debt, could have been generated elsewhere in the market—often more productively.

If, instead, a portion of the debt were held by foreigners, the picture would look different. Domestic capital could remain employed in domestic production, while foreign capital financed government borrowing. That does not eliminate the fiscal cost, but it does reduce the economic burden.

Further, it would also mean a reallocation of resources from producers to nonproducers. Or as Smith states:

the interest… is paid by industrious people, and given to support idle people.

The issue is not where the money ends up, but how it is raised and what it displaces.

Fallacy #2: Public Debt as “Capital”

Closely related is the idea that public debt represents an addition to national capital.

Smith dismantles this argument by tracing the origin of that “capital”:

the capital… advanced to government was… a certain portion of the annual produce turned away from serving in the function of a capital to serve in that of a revenue…

The act of creating public debt does not create capital. It consumes it.

From the perspective of the individual lender, capital may appear to be preserved or even increased—especially if they can sell the government security or borrow against it. But at the level of the economy, nothing new has been created. Resources have simply been diverted.

Had they not advanced this capital to government, there would have been… two capitals… instead of one…

Public debt is not a pool of wealth—it reflects capital that has already been consumed, often without generating future returns.

Fallacy #3: If We Can Bear This Debt, We Can Bear More

These arguments help justify the existence of public debt. But they also give rise to a third, more dangerous belief: that because debt can be sustained, it can be expanded without limit.

Because government debt can be bought and sold, and because there is a willing class of lenders, it can appear sustainable—even as it grows.

Smith notes that “Great Britain seems to support with ease a burden which, half a century ago, nobody believed her capable of supporting,” but warns against concluding that it can bear any burden.

Part of this apparent sustainability comes from the incentives of lenders. As Smith observes, government debt is attractive not only for its returns, but because it can be resold—often at a profit:

The security which [the state] grants to the original creditor is made transferable to any other creditor… and… generally sells in the market for more than was originally paid for it… Hence the inclination or willingness… to lend.

Today, this dynamic is reinforced by modern financial institutions. What is often described as an “insatiable demand” for government debt is, in part, sustained by central banks that act as backstops in secondary debt markets and ensure liquidity.

The presence of demand does not mean the absence of cost. It simply makes that cost easier to ignore.

Closing: What Smith Saw Clearly

Taken together, these arguments form a coherent warning—one that stands somewhat apart from the rest of The Wealth of Nations.

Across this series, a few themes have come up repeatedly.

In earlier entries on mercantilism, Smith showed how concentrated interests distort policy for their own benefit. In the discussion of colonies, he demonstrated how institutional differences—particularly around land, taxation, and trade—shape economic outcomes. And in his treatment of taxation, he emphasized principles of transparency, proportionality, and restraint.

This final chapter pulls those threads together.

Public debt, as Smith describes it, is not just a financial instrument. It is a political tool—one that allows governments to spend without immediately taxing, to pursue policies that might otherwise face resistance, and to shift burdens across time.

What makes Smith so striking here is not simply that he criticizes debt. It is that he understands the system that sustains it: the incentives of politicians, the willingness of lenders, and the imperfect visibility of costs to the public.

That combination—more than any single policy—is what makes persistent borrowing possible.

And it is why, more than two centuries later, his warnings are not just relevant—they are recognizable.

Now a new comment on Hickel et al.’s BMJ Global Health paper further weakens the original argument. In “Comment on: Structural adjustment: damages, reparations and pathways to non-recurrence,” Justin Callais, Vincent Geloso, Christian Lessmann and Joseph Steinberg take aim at the paper’s core empirical claims. Hickel’s paper criticized structural adjustment programs (SAPs), which were implemented by organizations such as the World Bank and the International Monetary Fund to help developing countries stabilize their economies and reduce their debt.

Callais and his coauthors find, however, that the evidence on SAPs is mixed and heterogeneous, and Hickel et al. do not provide a credible basis for sweeping claims that these programs broadly caused collapsing wages, rising poverty and worsening health across the Global South.

That matters, because the Hickel paper has already begun circulating as if it settled the debate. It did not.

The first problem with Hickel’s critique of SAPs is one of timing and aggregation. Hickel and his coauthors present regional averages, especially for sub-Saharan Africa, and treat the time period around 1980 as if it were the common moment when structural adjustment began. But that is not how the underlying history actually unfolded. Countries entered these programs at different times, under different conditions and with very different institutional settings. Compressing all those factors into a single regional narrative obscures more than it reveals.

Once the data are disaggregated to the country level, Hickel’s simple morality tale begins to fall apart, as shown above in figure 1 from Callais et al. Some countries experienced severe economic declines in the 1980s without ever entering SAPs. Others received such programs and later saw substantial improvements in living standards. That does not prove the programs were beneficial. But it does show that the claim “SAPs caused the regional collapse” is far too crude to survive contact with the underlying variation.

This is a recurring problem in bad empirical work: averaging across heterogeneous countries and then treating the average pattern as causal evidence. Regional aggregates are descriptive, not a way of identifying causal effects.

The second and even deeper problem is that correlation is not causation. Countries do not enter IMF and World Bank programs at random. They typically do so when they are already in severe economic distress: fiscal crisis, balance-of-payments pressure, debt problems, inflation, external shocks or political instability. In other words, the countries that receive these programs are often already on a downward trajectory, as figure B.4 from Callais et al. demonstrates.

That creates an obvious selection problem. If a country is spiraling economically and then enters a structural adjustment program, one cannot simply compare outcomes before and after and declare the program responsible for the deterioration. The deterioration may have been the very reason the program was adopted in the first place.

And the broader global context makes it even harder to tell what caused what. Many developing countries in the 1980s were hit by oil shocks, global recession, collapsing export prices, rising interest rates following U.S. monetary tightening, debt crises, conflicts, droughts and in some cases the spread of HIV. A credible causal analysis would need to disentangle these competing forces carefully; Hickel et al. do not provide that analysis. As Callais, Geloso, Lessmann and Steinberg note, repeated and staggered treatment timing, selection into treatment and heterogeneity across programs make clean causal inference especially difficult here.

This is why simple “before and after” storytelling is not enough. It is also why dramatic visual claims should be treated with caution. A graph can be rhetorically powerful yet methodologically empty.

The third major issue is conceptual: Structural adjustment is not the same thing as market liberalization. Hickel and his coauthors repeatedly treat SAPs as though they were equivalent to “neoliberal shock therapy” or free-market reform. But this is a serious conflation. Participation in an SAP does not mean a country actually implemented liberalizing reforms in any meaningful sense.

Countries’ compliance with IMF and World Bank conditions was often weak, partial, delayed, waived or renegotiated. In many cases, the formal existence of a program tells us little about whether privatization, deregulation, trade liberalization, fiscal consolidation or credible property-rights reform actually occurred. So even if one could establish some causal effect of SAP participation, that would still not establish a causal effect of genuine economic liberalization.

That distinction matters because the literature on actual liberalization using more credible methods generally points in the opposite direction from Hickel’s thesis. As I discussed in my earlier piece, studies using synthetic controls, matching methods and related approaches tend to find that liberalization raises income and improves a range of welfare outcomes. If many SAPs involved limited or failed reform, then poor outcomes associated with them cannot be straightforwardly blamed on economic freedom.

This is exactly why critics of markets so often slide carelessly between categories. They move from “an IMF program existed” to “free markets were imposed” to “free markets caused suffering.” But each step in that argument requires proof, and in this case the proof is missing.

The Callais et al. comment also raises further concerns about data quality and interpretation. Some of Hickel et al.’s claims appear to rely on nonstandard sources, factual mistakes and questionable readings of published figures. One especially striking example concerns wage declines in Brazil and Colombia. Hickel and coauthors reportedly cite enormous wage collapses that should immediately raise eyebrows. But the underlying source reveals that the numbers come from an index of legal minimum wages, not average wages, and the index values appear to have been misread as percentage declines. That is the kind of basic numerical error that should make readers much more skeptical of the larger argument.

When a paper offers headline-grabbing claims that align neatly with an ideological narrative, the burden of proof should rise, not fall. Extraordinary claims about millions of people being impoverished or killed by “neoliberalism” require extraordinary empirical care. They cannot rest on loose aggregation, bad counterfactuals, selection bias and category confusion.

None of this proves that every structural adjustment program was beneficial. The more defensible conclusion is the one Callais and his coauthors actually draw: The evidence is mixed, heterogeneous and context-dependent. Some programs may have failed. Some may have done harm. Others may have helped stabilize economies or paved the way for later improvements. The reality is varied, not monolithic.

Once we separate failed or weakly implemented adjustment programs from genuine liberalization, the broader record of economic freedom remains what it has long been: overwhelmingly favorable. Countries that secure property rights, reduce arbitrary state interference, open trade, stabilize money and permit markets to function tend to do better, not worse. That does not mean reform is automatic or costless. It means that the serious empirical question is not whether economic freedom is generally good for development, but which institutional and political conditions allow its benefits to emerge most fully.

So the new comment paper does more than identify flaws in one viral BMJ article. It reveals a familiar pattern in anti-market scholarship: Start with an ideological conclusion, use broad aggregates and weak causal reasoning to dramatize it, then smuggle in the claim that crises associated with debt, corruption, conflict or state failure were really the fault of “capitalism.” Once the data are examined carefully, the argument looks far less impressive.

The Hickel paper should not be treated as a decisive indictment of structural adjustment, much less of economic freedom. At most, it is an example of how not to make that case, and the larger lesson remains intact. Economic freedom is not the cause of mass poverty in the developing world. More often, it is one of the few durable routes out of it.

A new article by Vincent Geloso, Aleksander Psurek and Ashruta Acharya does exactly that. In “The Hole in the Doughnut: Formalizing and Testing a Key Model of Degrowth,” the authors take the most influential framework in the degrowth literature, Kate Raworth’s “doughnut model,” and formalize it, derive its empirical implications and test whether it actually works.

From Metaphor to Measurement

Raworth’s doughnut, shown in the figure below, has become the defining visual of the degrowth movement. It posits that economies should operate within a “safe and just space” bounded by a social foundation (meeting human needs) and an ecological ceiling (respecting planetary limits). The implication is that modern capitalism pushes societies outside this zone, producing both inequality and environmental degradation.

The appeal of the framework is intuitive, visually compelling and normatively attractive. Moreover, it’s not just a fringe idea in the economic and growth field—as table 1 of the Geloso et al. paper shows, it is one of the most widely cited degrowth models, gaining hundreds of new citations every year.

But until now, Raworth’s framework has largely escaped rigorous empirical scrutiny.

Geloso and his coauthors translate this doughnut framework into a measurable model. Social and environmental outcomes are expressed as normalized indices (ranging from 0 to 1), and “imbalance” is captured using the coefficient of variation (CV) across these indicators. Once formalized, the doughnut model generates a clear, testable prediction:

If capitalism (or economic freedom) drives unsustainable development, then more capitalist economies should exhibit greater imbalance, that is, a higher coefficient of variation across social and ecological indicators.

The Core Finding: The Prediction Doesn’t Hold

Using multiple datasets, specifications and indicator sets—including the Human Development Index (HDI) published by the United Nations Development Programme, inequality-adjusted HDI, Yale University’s Environmental Performance Index and Raworth’s own preferred measures—the authors test whether more economically free (i.e., more “capitalist”) countries exhibit greater imbalance.

Across panel regressions and cross-sectional analyses, the relationship between economic freedom and social/ecological imbalance is either statistically insignificant or runs in the opposite direction. In other words, more capitalist economies are not systematically more “unbalanced” in the sense the doughnut model predicts. If anything, they tend to be more balanced.

This conclusion is not a fragile result. It survives different measures of human development, different environmental indices, different functional forms and different samples and data constraints.

Each and every time, the doughnut model does not generate empirical results that match its own predictions.

Worse for Degrowth: Capitalism Improves the Levels Too

There’s an even more uncomfortable result for degrowth proponents. Not only does economic freedom fail to increase imbalance, it is often positively associated with higher levels of human well-being and environmental performance.

The authors explicitly note that economic freedom is “frequently associated with better outcomes” across measures like HDI, inequality-adjusted HDI and broader multidimensional indicators.

So, the data suggest two things simultaneously:

1. Capitalism does not create the tradeoffs the doughnut model predicts.

2. Capitalism is often associated with improvements across the very dimensions the doughnut cares about.

These findings directly contradict the central narrative of degrowth, that economic growth systematically undermines social and ecological outcomes.

This paper is significant not because it “debunks” every concern about growth or the environment. Rather, it shows that the leading theoretical framework of degrowth fails its most basic test: internal empirical consistency.

As the authors put it, before we argue about policy or ideology, we should ask whether the theory “can walk.” If it cannot generate predictions that match reality, there is little reason to build policy on top of it.

And that’s exactly the problem here. The doughnut model claims that growth and capitalism inherently produce imbalance. But when translated into measurable terms, that claim simply does not hold.

The Real Takeaway: Degrowth Adds No Explanatory Power

One of the most interesting aspects of Geloso et al.’s paper is what it doesn’t claim. The authors are careful not to argue that markets solve all environmental problems. Nor do they deny the existence of externalities, tradeoffs or policy challenges.

Instead, they make a narrower point: Everything the doughnut model tries to explain can already be explained within standard economic frameworks.

Whether you favor property rights and market-based solutions, Pigouvian taxes and regulation, or some hybrid approach, all these perspectives can account for the empirical patterns we observe.

Degrowth, by contrast, adds no explanatory power. It does not improve our understanding of the problem, and it fails to generate correct predictions. As the authors conclude, the model is “not empirically supported and may provide a limited guide for policy.”

The Doughnut Is Stale

Degrowth has gained influence not because it has been rigorously validated, but because it offers a compelling narrative: that growth is the root cause of modern crises. This new paper challenges that narrative at its foundation.

If the most prominent model of degrowth cannot survive empirical testing, even when evaluated on its own terms, then the intellectual case for degrowth becomes much weaker.

The debate over growth, sustainability and human well-being is not going away. But after this paper has tested the core claims of the degrowth literature, it is harder to argue that degrowth provides a serious analytical framework for that debate.

Here are important numbers highlighted in the report: in fiscal year 2025, Americans 65 and older received $43,700 per person in federal spending. Americans under 26 received $4,300. Workers between 26 and 64, the people currently financing the system through their payroll Taxes, income taxes, and labor received $7,300.

Before anyone jumps to their preferred political explanation, let me say now that the solution isn’t to raise taxes or spend more on younger Americans. The government is way too big and inefficient as it already is. But more on that later.

What the Numbers Actually Show

Of the $7 trillion the federal government spent in fiscal year 2025, about $2.6 trillion went to things that cannot be assigned to any particular age group. That includes defense, interest on the debt, and a few other things. The remaining $4.4 trillion went to identifiable beneficiaries.

Of that $4.4 trillion in direct age-assignable spending:

• Children and young adults under 26 received $449 billion, or 10 percent

• Working-age adults received $1.2 trillion, or 28 percent

• Retirees received $2.7 trillion, or 62 percent

Seniors receive 10 times as much per capita from the federal government as children and young adults, and 6 times as much as working adults. This imbalance is expected to grow.

Two programs drive the retiree dominance almost entirely: Social Security ($1.3 trillion to older adults) and Medicare ($835 billion). Together, they account for 80 percent of all federal spending on people 65 and older.

These programs are the drivers of our current and future spending and debt. The United States government is often described as a military superpower that happens to run a social safety net on the side. That description is now outdated. A more accurate one: the federal government is a retirement home that maintains an army for security, collects taxes from the staff to pay the residents’ bills, and sends the remaining expenses to the residents’ grandchildren on credit.

Penn Wharton’s political economy analysis offers a sobering explanation for why this persists. Old people vote, and middle-aged voters can imagine benefiting from the benefits.

That democratic support doesn’t make it okay, especially since, as designed, it is also an immoral transfer of wealth from the relatively young and poor to the relatively old and wealthy. And older people can’t even really claim that they have paid for the benefits entirely, as they often imply.

Did Seniors Actually Pay for This?

This is the question that generates the most heat and the least light in public debate. The honest answer is: partly, but significantly less than most people assume, and the shortfall grows every year.

It is true that today’s retirees spent careers paying payroll taxes. But the pay-as-you-go structure of Social Security and Medicare Part A means those past contributions were not saved or invested on the contributor’s behalf. They were transferred immediately to whoever was retired at the time. The implicit contract was always: your taxes fund today’s retirees, and tomorrow’s workers will fund you. What that contract depends on and what is now strained is a favorable ratio of workers to retirees.

That ratio has deteriorated sharply. When Social Security was designed, there were many workers per retiree. Today, the ratio is roughly 2.7 to 1 and falling. The math that made the original bargain viable no longer holds.

Medicare compounds the problem. Nearly 45 percent of Medicare( Parts B and D) is funded through general revenues, not payroll taxes at all. Beneficiaries paid no dedicated tax for a large portion of what they now receive. And across both Social Security and Medicare, the Penn Wharton analysis notes that many past and current retirees have received or will receive more in present-value benefits than they contributed through payroll taxes.

Two Wrong Answers

The data is stark enough that it will inevitably be recruited into arguments for solutions that do not actually follow from it. Two in particular deserve pushback.

The first wrong answer is: More taxes

The fiscal arithmetic does not support this as a primary solution. The federal government’s structural imbalance is driven by the growth of age-based entitlement spending, not by an insufficient top marginal tax rate. The Congressional Budget Office projects that Social Security and Medicare alone will account for essentially all deficit spending over the next three decades. Closing that gap through income taxes on high earners would require rate increases of a magnitude that would have significant effects on investment, capital formation, and economic growth, and still would not fully close the structural shortfall, because the problem is not the revenue baseline; it is the spending trajectory. Redistribution from the rich to the poor isn’t a substitute for entitlement reform. And as I have said before, if Congress doesn’t want to reform these programs and borrow instead, then inflation will take care of the adjustment.

The second wrong answer is: the solution is to spend more on children and young adults.

The Penn Wharton data showing that children receive far less federal spending than seniors is sometimes read as an argument for expanding programs aimed at younger Americans. That reading misses the point. The problem is not that child-directed spending is too low in absolute terms (I can point to many programs that shouldn’t exist or that are distortive and counterproductive when directed to children now); it is that age-based entitlement spending is on a trajectory that will crowd out everything else. Adding new spending commitments on top of an already-unsustainable baseline does not solve the structural imbalance; it deepens it.

The Penn Wharton analysis is another clear-eyed accounting of where the federal government’s resources actually go and a demonstration that current trends are not politically stable, demographically sustainable, or fiscally viable.

In Book V, Adam Smith turns from public works to another core function of government: taxation. If governments must raise revenue, the question is not whether to tax, but how.

Smith’s answer is both systematic and modern. Taxes shape behavior and alter decisions, making them as much a problem of incentives as of revenue collection. And badly designed taxes can do far more harm than the revenue they raise.

At the same time, his analysis is not without its limits. While he identifies many of the distortions taxes create, he does not fully account for how their burden is shared across individuals.

The Four Maxims of Taxation

Smith begins, unsurprisingly, with an enduring framework for thinking about taxation: his four maxims.

The first is that taxation should be proportional to ability:

The subjects of every state ought to contribute towards the support of the government… in proportion to the revenue which they respectively enjoy under the protection of the state.

It becomes clear later that Smith’s emphasis is less on proportionality than on participation, noting that he finds it “not very unreasonable that the rich” should pay a larger proportion.

The second maxim is certainty:

The tax which each individual is bound to pay ought to be certain, and not arbitrary. The time of payment, the manner of payment, the quantity to be paid, ought all to be clear and plain to the contributor, and to every other person.

The emphasis here is less on convenience and more on limiting opportunities for abuse by tax collectors, who could otherwise extract payments beyond what the law requires.

We don’t worry much about tax collectors trying to line their pockets in the U.S., but arbitrariness is a major issue here nonetheless, particularly the income tax.

Individual tax filers are responsible for calculating their income tax each year, often having multiple acceptable ways to do so. Utilizing additional methods to lower your tax bill, such as credits and itemized deductions, increases both the complexity of your tax return and the risk of being audited by the Internal Revenue Service—which also does not know what you are legally required to pay—even when you’ve followed all applicable tax laws.

The third maxim is convenience:

Every tax ought to be levied at the time… in which it is most likely to be convenient for the contributor to pay…

And the fourth maxim is efficiency:

Every tax ought to be so contrived as both to take out and to keep out of the pockets of the people as little as possible over and above what it brings into the public treasury of the state.

This is the most sophisticated of Smith’s maxims. Taxes impose costs beyond what they raise in revenue—costs for administration and collection, discouraging productive activities, resources used on attempted evasion, and the inconveniences from enforcement. A well-designed tax minimizes these additional burdens.

The Laffer Smithian Curve

By now, Smith’s incredible perceptiveness is more than obvious. But to highlight another unique insight he made almost 200 years before it became mainstream, see the following quote:

High taxes, sometimes by diminishing the consumption of the taxed commodities, and sometimes by encouraging smuggling, frequently afford a smaller revenue to government than what might be drawn from more moderate taxes.

When the diminution of revenue is the effect of the diminution of consumption there can be but one remedy, and that is the lowering of the tax.

In other words, Smith proposed an 18^th century version of the Laffer curve—the idea that higher tax rates discourage activity (and encourage evasion), leading to lower revenues.

Why Property Is Taxed

Like many others, I’ve asked myself why the government taxes property, particularly land. “Is it really your property if the government taxes it?”

Smith offers a practical explanation for why taxes on property arose. Unlike capital or income, land is visible, immobile, and consistent:

the quantity and value of the land which any man possesses can never be a secret, and can always be ascertained with great exactness. But the whole amount of the capital stock which he possesses is almost always a secret, and can scarce ever be ascertained with tolerable exactness. It is liable, besides, to almost continual variations… Secondly, land is a subject which cannot be removed; whereas stock easily may. The proprietor… would be apt to abandon the country in which he was exposed to a vexatious inquisition, in order to be assessed to a burdensome tax, and would remove his stock to some other country where he could either carry on his business, or enjoy his fortune more at his ease.

Thus, the development of taxes on land appears to have been driven less by theory than by administrative practicality. Governments taxed what they could observe, measure, and collect without excessive cost or resistance. Land, unlike capital, could not be hidden, moved, or easily revalued, making it a stable and reliable tax base.

But this also reveals a deeper point. Tax systems are not designed in a vacuum. They are constrained by what can be enforced. The prevalence of property taxes is not necessarily evidence of their efficiency, but of their administrability. What is easy to tax is often what gets taxed.

For all of Smith’s insight into taxation, however, his analysis is not complete.

Where Smith Falls Short

Smith falls short in one important area: tax incidence.

He clearly understands that the person legally responsible for a tax is not always the one who ultimately bears its burden. In discussing taxes on profits, he notes:

…the final payment of the tax… will fall upon the consumer…

And in discussing taxes on necessities, he argues that they raise wages and are ultimately passed on through prices.

But he tends to treat tax incidence as shifting entirely from one group to another—producers to consumers, or employers to landlords. What he does not fully consider is that the burden may be shared.

In reality, taxes are often distributed across multiple parties:

• partially absorbed by producers

• partially passed on to consumers

• and partially reflected in wages or returns to capital

This matters because it complicates the analysis of who truly pays.

Smith’s framework captures the direction of these effects, but not their full distribution.

The Limits of Tax Design

Smith’s analysis of taxation is remarkably modern. He understood that taxes are not neutral—they shape behavior, alter incentives, and impose costs beyond what they raise in revenue. His maxims remain a useful guide: taxes should be certain, convenient, proportional, and efficient.

He also recognized that poorly designed taxes can backfire. High rates encourage evasion. Arbitrary systems invite abuse. Taxes imposed without regard to incentives can reduce both economic activity and government revenue.

But while he understood that taxes can be shifted, he did not fully account for how their burden is shared. In practice, taxes rarely fall entirely on one group. They are distributed across producers, consumers, workers, and owners of capital in ways that are often difficult to observe.

This matters because it complicates the question of who ultimately pays.

Tax policy is not simply about raising revenue from a defined group. It is about understanding how that burden spreads through the economy.

And it is in that spread—often unseen—that many of the real effects of taxation are found.

Despite the many errors with this theory, it has diverted our attention from a real and more fundamental problem now facing the US. A new paper from economists at Harvard, the Federal Reserve, and Columbia points to some bigger issues.

Two Verdicts, Rendered

The new paper tracks something called the Treasury convenience yield. That is the premium that global investors are willing to pay to hold US government bonds rather than equivalent bonds from other countries such as Germany, Australia, or Sweden. When this premium is high and positive, America’s debt is genuinely special. Global investors want US bonds so badly that they will accept lower returns to own them. This specialness is a concrete financial benefit of the US dollar’s reserve-currency status; it is one factor that has allowed the US government to borrow more cheaply than anyone else.

That premium has largely disappeared. For ten-year maturities it turned negative around 2010. For five-year maturities, it turned negative around 2012. Even at much shorter maturities it has turned negative since 2023. Across the entire maturity spectrum, US Treasuries are no longer inherently more special than equivalent bonds from other stable developed countries, a finding the paper describes as a pronounced secular decline.

A reasonable objection here is that convenience yields are also shaped by factors beyond any one country’s fiscal position, including changes in bank liquidity regulations, collateral rules in derivatives markets, and shifts in how central banks manage reserves. I take those measurement complications seriously. But a fifteen-year decline across the entire maturity spectrum, correlated with rising debt-to-GDP across administrations of both parties, is not a story that institutional plumbing can fully explain. The trend is telling us something, and it is not good.

While Treasury convenience has collapsed, dollar convenience has done the opposite. The premium that market participants pay to hold dollar-denominated risk-free assets remains strong and actually rises during periods of global stress. The dollar’s role as the world’s dominant funding currency has not changed. When panic hits, the world still rushes into dollars. That has not changed.

The paper documents a striking decoupling: the same global markets are rendering two separate verdicts at the same time. The dollar retains its special and safe-haven characteristics. The Treasury bond, increasingly, does not.

The paper’s explanation for this divergence is technical but important. Dollar convenience reflects the structural role of the US dollar in global funding markets: foreign institutions need dollars for hedging and funding, global banks earn an intermediation fee supplying those dollars through the FX swap market, and this demand is sustained by how the international financial system is built. Treasury convenience, by contrast, reflects the relative supply of US government bonds compared to government bonds issued by other developed countries. As the US has issued far more debt relative to GDP than its peers, the convenience premium on Treasuries has been progressively crowded out. The two measures are driven by different forces.

One interpretation of this divergence is that global markets are distinguishing between two different things when they look at the United States. The dollar’s ongoing strength reflects the structural centrality of American financial markets to global commerce and funding. The erosion of Treasury convenience reflects something narrower and more troubling: the US government has issued so much debt, relative to every other developed country, that its bonds no longer command a premium. Those are not the same judgment, and it matters enormously that they are now pointing in opposite directions.

About Liberation Day

For those interested in what the bond-market data show about “Liberation Day” specifically, the paper provides a live test and the results are instructive. In the ten days after the tariff announcement, dollar convenience rose, consistent with the dollar’s historical safe haven behavior. Treasury convenience fell sharply at medium and long maturities. The countries best positioned to substitute away from American bonds, those with low debt and strong institutions like Germany and Australia, moved most aggressively. It was so bad that the most aggressive tariffs were paused in response to market pressure.

The falling Treasury convenience yield was the market’s simultaneous verdict on the government’s side of the ledger.

The Indictment of Our Political Class

When the paper goes looking for why Treasury convenience has collapsed, the answer is uncomfortable and bipartisan. It is not China’s fault. It is not the fault of international monetary system. it is the relative volume of US government debt compared to other developed economies. Every time the US debt-to-GDP ratio rises by one percent, Treasury convenience falls by a measurable, statistically significant amount. The relationship holds across time periods, across maturities, across administrations of both parties.

The Obama administration ran large budget deficits through the financial crisis and never fully normalized them. The Trump first term jacked up the deficit during a strong economy when surpluses should have been accumulating. The Biden administration spent enormously through COVID and added long-term commitments on top. The second Trump administration is now continuing down this path at faster speed. Each wave of borrowing pushed the convenience yield further negative, in a relationship the paper quantifies.

As Treasury convenience falls, borrowing costs rise. Rising borrowing costs expand the budget deficit directly through higher interest payments. A larger deficit requires that more debt be issued. More debt further suppresses the convenience yield. The US is not approaching a fiscal cliff so much as it’s in a fiscal downward spiral, and the speed of descent is not linear.

There’s a painful irony in all of this. The very strength of the American private economy has made the fiscal deterioration easier to sustain and easier to ignore. A weaker private economy would have produced a crisis much sooner; bond markets would have forced discipline through rising yields and collapsing demand long ago. Because the dollar remains essential to global finance and the American economy remains genuinely impressive, the US has been able to borrow far beyond what would have been sustainable for any other country. The private sector’s strength bought the government time. The government used that time to borrow more rather than restore discipline – actions that, in a terrible irony, threaten to inflict serious damage on the very private economy that allowed the government to behave so recklessly.

The Conversation We Refuse to Have

The federal budget is Social Security, Medicare, Medicaid, and interest payments. Everything else is noise. DOGE discovered this reality the hard way. Total federal spending rose by $300 billion in fiscal year 2025 above its original baseline despite the loud and garish theater of contract cancellations and workforce reductions, because cutting consultants and foreign aid while leaving entitlements untouched is like bailing out a flooding large ship with a teaspoon.

The exorbitant privilege, America’s ability to borrow cheaply because its debt instruments were the world’s safest assets, has been spent down by the government. The cost of continued denial is now showing up in the prices of the bonds that finance that denial. At some point the markets will stop waiting for Washington to notice.

The cost of T+X settlement, when X is expressed in days:

In our digital world, where information travels at 186,000 miles per second, it is strange that the unit of account used to express the duration of the trade settlement cycle is not seconds, minutes or hours but entire days after the trade date (T). Settlement took T+5 days in the eighties, T+3 in the nineties. Some regions of the world still function at T+2 today, such as the European Union (EU), but all are converging to a T+1 delay (the official U.S. delay since May 2024 for most transactions). Yes, there is clear progress over time, but in an environment where prices are updated every thousandth of a second, the frictions generated by a single trade remain high.

First, a trade that is not finalized locks the asset and the capital used to buy it. The seller cannot use the cash for other opportunities, and the buyer cannot resell the stock right away. If, in practice, investors find workarounds so they can prefund trades or presell assets, the workarounds all generate costs. Meanwhile, cash and assets are locked away from market opportunities as long as the trade isn’t cleared.

Also, the longer a trade takes to settle, the longer both buyers and sellers are exposed to the possibility that their trading partners might default before the exchange is finalized. This risk is made worse because the price of the traded asset may vary during that time. These counterparty and market risks justify consequent margin requirements by clearinghouses: more cash being locked away from investors, forced to stay idle while the trade is being confirmed. Shortening the settlement cycle not only reduces the delay during which all this capital is locked, but also lowers the required amount to begin with as the market and counterparty risks are reduced.

Measured effects:

The National Securities Clearing Corporation (NSCC), the CCP for the U.S. market—managing risk by guaranteeing trades and maintaining a clearing fund to cover potential defaults—monitored a 30% decrease in its clearing fund ($3.7 billion USD) after the U.S. moved to T+1 settlement. The European Securities and Markets Authority (ESMA), the EU securities regulator, foresees a proportionally bigger decrease of 42% (representing 2.4 billion EUR) if and when the EU switches to T+1.

The E.U.’s inability to move to a T+1 settlement cycle doesn’t just impose opportunity cost. The switch of its most important counterparties to a T+1 settlement cycle induced a misalignment tax: The EU market has to manage securities both in T+2 and T+1 for its cross-border activity, thereby increasing administrative costs. Moreover, an important liquidity mismatch is generated: When an EU-based fund buys U.S. securities, it must provide the cash by T+1, while the settlement of the investment fund’s own shares in the EU still occurs by T+2. This gap of at least one day forces funds to secure extra cash to cover the timing difference. The gap is worsened on Thursdays because the fund must settle with the U.S. on Friday while not receiving its subscription or redemption proceeds before the following Monday. ESMA consequently measured a decline in liquidity of EU exchange-traded funds invested in U.S. equities following the U.S. progress to T+1.

Although the EU led the U.S. by three years in the move to T+2 (2014 versus 2017), it is now lagging in T+1 implementation. This delay persists despite the known benefits of shorter cycles and the cost of market misalignment.

Comparing the quick U.S. switch to T+1 with the lagging EU approach demonstrates a stark difference. The EU process is riddled with administrative burden, incessant impact assessing and task force assembling.

Quick overview of the U.S. process:

The 2021 GameStop short squeeze revealed how the T+2 settlement cycle creates bottlenecks during periods of extreme volatility. During the event, the price of GameStop (GME) rose nearly 1,000% in a single week, dramatically increasing the “volatility component” of the risk that clearinghouses had to manage.

Because two days of trading remained unsettled at any given time, the NSCC was exposed to a massive amount of counterparty risk. To mitigate it, the NSCC issued emergency margin calls to member firms. Robinhood alone was required to post an additional $3 billion in collateral in a single morning. Because many retail brokerages lacked the immediate liquidity to meet these multi-billion-dollar demands, they were forced to restrict their customers from buying the stock, which effectively halted market activity to prevent a systemic default.

This crisis triggered an immediate reaction from U.S. institutions seeing that the T+2 settlement cycle was a source of systemic fragility. By February 2021, the Depository Trust & Clearning Corporation (DTCC) had published a white paper arguing in favor of shortening the cycle. It quickly became clear to most market participants that cutting the settlement time in half was the most effective way to lower the risk per trade across the entire ecosystem.

Following an April 2021 partnership between the DTCC, the Securities Industry and Financial Markets Association, and the Investment Company Institute that created a technical road map by December of that same year, the SEC formally proposed the rule change in February 2022, and it was adopted by February 2023. A year-long testing phase culminated in the system-wide implementation of T+1 on May 28, 2024. It took three years.

Quick overview of the EU process:

Since the beginning of the 2020s, the idea of shortening the EU’s settlement cycle was present in different reports, but the trigger point for any kind of institutional attention was the U.S. announcement of its switch. In September 2022 the Association for Financial Markets in Europe released a first white paper dedicated to T+1 settlement. Almost a year later, ESMA launched a call for evidence around T+1. In November 2024, ESMA finally conceded the necessity of a switch. In February 2025, the EU Commission drafted a regulation proposal. The proposal then had to be negotiated by the European Council and parliament, which led to a lightly amended provisional agreement. In September 2025, both institutions definitively voted in favor of the regulation. Officially published on 14 October 2025, the regulation itself imposes a deferred entry into force on 11 October 2027. It took seven years.

It should be noted that this deferred entry into force was mandated by the private sector, in the name of precaution, to ensure a smooth transition. More realistically, incumbents were reluctant to pay the investment costs of a T+1 switch and feared the risk of competition during the implementation process. Unsurprisingly, established firms fear being displaced by new, tech-driven actors whose ability to implement T+1 more efficiently could reshuffle the market’s competitive landscape. This postponement basically grants a risk-free, comfortable innovation timeline for incumbents and discourages smaller, younger entrants who lose the incentive to innovate, the regulation ensuring that larger players will have reached technological parity before the market officially opens.

In such an institutionalized system, business appears more like a rent distribution negotiation with EU institutions than an actual competition between service providers. Sometimes, surely only sometimes, it does feel like the precaution principle isn’t so much for consumer protection as for incumbent comfort.

Does late for T+1 necessarily mean late for T+0?

This waste of time is too costly. With global T+0 settlement already on its way, the EU cannot afford to be late again.

It should be noted that in 2015, the European Central Bank launched its Target 2 securities platform (T2S), a common infrastructure already capable of settling cross-border trades with central bank money in real time. Despite this technical availability of T+0 for over a decade, T2S remains underutilized, notably due to persistent market fragmentation. Because participation is not harmonized, several EU member states continue to operate national legacy settlement systems in parallel, adding redundancy and compliance costs for participants (settlement is 2 to 4 times more expensive in the EU than in the U.S.). T+0-worthy infrastructure was deployed a decade ago, yet T+2 remains the standard.

The EU is a knowledgeable and capable body, fully aware of the stakes, advantages, drawbacks, challenges and solutions for an effective common capital market. But the task of reaching a 27-state consensus, combined with systemic risk aversion and an exhaustive consultation process with interest groups (often working as a mechanism for legacy actors to delay high-cost transitions and kill competition), paralyzes progress in such a way that even a politically noncontroversial reform like the T+1 switch has to take more than twice as long as it did in the U.S. This branded precaution costs.

The size of the tax increase is substantial. Once fully implemented, Washington’s combined state and local top marginal rate will exceed 18%. When combined with federal taxes, top earners will face marginal rates approaching 58%.

But the policy faces immediate legal and political uncertainty. For more than 90 years, the Washington Supreme Court has held that income constitutes property under the state constitution, meaning it must be taxed uniformly rather than at graduated rates. On that basis, former Attorney General Rob McKenna and the Citizen Action Defense Fund have already filed suit challenging the tax’s constitutionality. At the same time, opponents are pursuing a referendum to put repeal directly before voters.

Notably, the tax will not take effect until January 1, 2028, with the first payments due in April 2029. That delay gives high-income taxpayers years to adjust their behavior in advance, whether by shifting income, changing residency or restructuring how and when they realize earnings. As the chart below demonstrates, nearly 15,000 high-income individuals already left the state in the three years following the enactment of a capital gains tax. Static projections that ignore these responses are therefore especially unreliable.

Even setting all those challenges aside, there is a further reason for skepticism of the millionaires tax: The revenue projections themselves are almost certainly too optimistic.

One empirical guide we have comes from California’s Proposition 30. In a careful 2024 study using administrative tax data, Joshua Rauh et al. find that a relatively modest 3-percentage-point increase in top marginal rates, adopted in 2012, led to large behavioral responses among high earners. These responses—primarily changes in reported income rather than migration—eroded 45% of the proposition’s expected revenue within the first year and over 60% within two years.

At first glance, Washington’s policy might seem comparable. But on the economically relevant margin, its effects will be much larger.

What matters for behavior is not the statutory rate increase itself, but the change in the net-of-tax rate—the share of an additional dollar of income that taxpayers keep. On this margin, Washington’s new tax represents a shock roughly 2.8 times larger than California’s Proposition 30.

If a smaller tax increase in California wiped out more than half of projected revenues within two years, a tax increase nearly three times as large risks eroding an even greater share of the tax base. A simple extrapolation suggests that the Evergreen State could lose more than 100% of projected revenue. But that is precisely the point: The behavioral response implied by the evidence is large enough that it could nearly eliminate the expected revenue altogether.

The static estimates of revenues in the $3–4 billion range overlook a much broader problem. When high earners leave, the state does not just lose their income tax revenue. It also loses the sales taxes they pay, the property taxes they contribute and the business and occupation taxes tied to the firms they own and operate. Many of these individuals are also major employers and investors, supporting thousands of jobs and significant local economic activity. As their spending, investment and business activity shift elsewhere, the erosion of the tax base extends well beyond the income tax itself.

There is one important nuance worth acknowledging. California already had a high state income tax prior to Proposition 30, whereas Washington is introducing this tax from a base of zero. That difference could cut in either direction. On one hand, the introduction of a new, high marginal rate may represent a larger behavioral shock. On the other hand, California’s taxpayers may already have had more established mechanisms for income shifting and tax planning. Even so, the core lesson from the evidence remains: High-income taxpayers are highly responsive to marginal tax rates.

The broader takeaway is straightforward. Ignoring behavioral responses leads to systematically inflated revenue forecasts, particularly when targeting a small, highly mobile group of top earners. If this “Millionaires Tax” holds up to legal challenges, Washington is about to test that lesson in real time.

This is part of a broader series we’re working on. Like many on the free market side, we welcome the renewed focus on abundance. It is encouraging to see parts of the left rediscover something economists have long understood, namely that overregulation makes it harder to build, invest, and grow.

But the conversation remains incomplete. As we noted a few months ago, the current debate over abundance focuses almost entirely on regulation. It says very little about the tax code, which often works in the same direction, quietly undermining supply, investment, and mobility. Our latest piece focuses on housing, but the problem extends more broadly to infrastructure and industrial construction.

That blind spot matters because the tax code is not neutral. Jack and I have talked about that a bunch of places, but see here. It does not simply raise revenue. It shapes behavior. In housing, it does so in two especially damaging ways.

First, it freezes the existing housing stock.

The combination of demand-side subsidies and capital gains taxes discourages turnover. Policies like the mortgage interest deduction push prices up without increasing supply. At the same time, capital gains taxes create a lock-in effect that discourages people from selling homes that no longer fit their needs.

The result is misallocation. Larger homes sit underused. Younger families struggle to buy. Housing doesn’t move to its highest-value use.

Second, and more broadly, the tax code discourages building, whether housing, infrastructure, or industrial capacity.

While policymakers have improved the treatment of equipment, most structures are still penalized through long depreciation schedules. That applies not just to apartment buildings, but to office space, energy infrastructure, and industrial facilities.

“The result is a higher after-tax cost of building and an additional tax on structures relative to other investments.”

This is a systematic bias against building things. It lowers the return to long-lived capital and makes fewer projects viable. In other words, even if you fix permitting, you are still taxing construction more heavily than other forms of investment.

The fixes follow the diagnosis. The two problems require two different solutions. For the existing housing stock, the ideal fix is to eliminate the capital gains tax altogether and move to a consumption-based tax that stops penalizing saving and investment. Incremental steps such as indexing capital gains for inflation, expanding the primary residence exclusion, and relaxing 1031 exchange rules would help at the margins, though each requires careful design to avoid creating new distortions. For new construction, the priority is straightforward: extend the full immediate expensing that equipment now enjoys under the Big Beautiful Bill to all structures, residential and commercial alike. That single change would reduce the after-tax cost of building more meaningfully than most permitting reforms currently under discussion.

Over at his Substack, Adam rightly notes that some of the reforms are prime for being included in Reconciliation 2.0. He adds a few ideas about how to pay for them.

The broader point is simple: If you are serious about abundance in housing and construction in general, you cannot focus only on zoning and permitting. You also have to look at the tax base. Right now, the tax code rewards holding assets over reallocating them, consumption over investment, and incumbents over new entrants. That is the opposite of an abundance agenda.

In Book V, Adam Smith turns to one of the few areas where he appears to grant a clear role for government. Alongside national defense and the administration of justice, he includes a third duty: the provision of certain public works.

Roads, bridges, canals, and harbors, he argues, are often necessary for facilitating commerce. Without them, the extent of the market is constrained.

He also addresses how these public works ought to be financed, stating that they should be paid for with user fees. But this raises a more difficult question. If these works can be funded and maintained by user fees, why are they the responsibility of government at all?

Smith’s answer is not always consistent. In some cases, he relies on private incentives. In others, he dismisses them too quickly. The result is a theory of public works that is more coherent in its principles than in its conclusions.

When Public Works Are Justified

Smith begins with a clear criterion for when public provision may be necessary. Certain works, he argues, are so costly that no individual—or even small group—could expect to recover their expense through profit:

The third and last duty of the sovereign or commonwealth is that of erecting and maintaining those public institutions and those public works, which, though they may be in the highest degree advantageous to a great society, are, however, of such a nature that the profit could never repay the expence to any individual or small number of individuals…

This is the core of his argument: public works are justified not simply because they are useful, but because they cannot be profitably supplied. The question, then, is whether that assumption is correct.

Smith then turns to a more concrete observation: as commerce expands, so too does the need for infrastructure to support it.

The expence of making and maintaining the public roads of any country must evidently increase with the annual produce of the land and labour of that country… The strength of a bridge must be suited to the number and weight of the carriages which are likely to pass over it.

Infrastructure, in this sense, is not arbitrary. It follows commerce. As the volume of trade increases, so does the demand for roads, bridges, and canals capable of sustaining it.

From this, Smith draws a narrower conclusion than is often assumed. These works are justified not because they are public, but because they reduce the cost of exchange. Their role is to expand the market by lowering the cost of moving goods within it.

Who Should Pay

While Smith allows for public works, he is far more restrictive about how they should be funded.

It does not seem necessary that the expence of those public works should be defrayed from that public revenue… The greater part of such public works may easily be so managed as to afford a particular revenue sufficient for defraying their own expence…

His principle is consistent: those who benefit should pay.

A highway… may… be made and maintained by a small toll… a harbour, by a moderate port-duty… The post-office… affords… a very considerable revenue to the sovereign.

These fees do more than raise revenue. They align payment with use. Those who impose the greatest wear and tear pay the most, and those who benefit most from the infrastructure bear its cost.

When the carriages… pay toll in proportion to their weight… they pay… in proportion to the wear and tear… It seems scarce possible to invent a more equitable way…

Just as importantly, user fees discipline investment. Works are built where they are needed and scaled to what commerce can sustain.

They can be made only where that commerce requires them… A great bridge cannot be thrown over a river at a place where nobody passes…

Where Smith’s Argument Breaks

But if these projects can be “made and maintained” through user fees, it raises a deeper question: why are they a duty of government at all?

Smith himself acknowledges that some of these works can be privately provided. Navigable canals, for instance:

In several different parts of Europe the ton or lock-duty upon a canal is the property of private persons, whose private interest obliges them to keep up the canal. If it is not kept in tolerable order, the navigation necessarily ceases altogether, and along with it the whole profit which they can make by the tolls.

In other words, even within his own framework, some “public works” can be reliably provided through private incentives.

But when he turns to roads, Smith abandons this logic. He argues that roads cannot “with any safety” be entrusted to private operators because they remain usable even when poorly maintained, allowing tolls to be collected without corresponding upkeep.

The concern is real—but it is not unique to private provision. It is a problem of weak competition and poorly defined incentives, not ownership itself.

Local Road Monopolies

Some background is in order.

Prior to the 17th century, Great Britain’s roads were maintained by local governments, known as parishes. As trade expanded, however, road quality became an issue, particularly with the use of large wagons. Financing mechanisms were limited: statute labor required unpaid work from residents, and property taxes were capped.

In response, Parliament authorized the creation of turnpike trusts—local entities with the authority to levy tolls and issue bonds. These trusts, often composed of local elites, effectively created a form of controlled monopoly over road provision.

Smith himself noted the resulting problems: high tolls, poor maintenance, and what he described as a “slovenly” approach to upkeep.

What Smith treats as a failure of private provision is more accurately a failure of competitive conditions—comparable to modern American utility monopolies.

A private system would differ in one crucial respect: it would allow for competition.

If a road proprietor charged high tolls while neglecting maintenance, adjacent landowners would have an incentive to build competing routes. Even if entry were costly, the possibility of competition would discipline pricing and quality.

More importantly, competition would not be limited to new entrants. Existing roads already competed along trade routes. Merchants, observing differences in tolls and road quality, would choose the most profitable paths. Poorly maintained roads would lose traffic; well-maintained roads would gain it. The proprietor of a well-maintained road would hear reports of poorly maintained roads elsewhere, alerting him to profit opportunities.

Public Works Without Public Provision

Smith’s discussion of public works is often read as a justification for government provision. But read more closely, it points to a different conclusion.

He consistently favors pricing tied to use, funding tied to beneficiaries, and administration tied to those with a stake in outcomes. These are not uniquely public principles. They are the same principles that govern successful private markets.

The tension in his argument arises when he departs from those principles. Where incentives are clear and competition is possible—as with canals—he trusts private provision. Where they appear weaker—as with roads—he defaults to public administration, without fully considering how institutions might restore those incentives.

The result is an incomplete argument. Smith identifies real coordination problems, but he does not fully explore how profit, property rights, and competition might solve them.

The question, then, is not whether public works are necessary. It is whether they must be public.

And in many cases—even in Smith’s own time—the answer was no.

It seems almost nonsensical that we still have to debate something we’ve long known to be true: Economic freedom improves living standards. Yet last week the medical journal BMJ published a new paper authored by anthropologist and de-growth proponent Jason Hickel and his colleagues. The authors describe the supposed harm caused by International Monetary Fund and World Bank structural adjustment programs in the Global South since the 1980s.

Without presenting any causal evidence, the paper suggests that privatization, deregulation and austerity led to negative impacts on wages, poverty, inequality, maternal mortality, infant mortality and healthcare access.

Notably, this isn’t the first paper in which Hickel has argued that increasing levels of economic freedom in the world led to declining living standards. He has argued in prior publications that China had lower rates of poverty than many market economies when it was a socialist planned economy, and that it was the shift toward economic freedom in the 1980s and 1990s that caused extreme poverty in China.

In 2023 he coauthored an article in which he argued that before the emergence of capitalism, extreme poverty had been virtually nonexistent or rare, while economic freedom was catastrophic for living standards of the vast majority of people in Europe and globally.

In 2024 sociology professor Tibor Rutar checked the robustness of these findings and analyzed additional sources and data that Hickel had not. Rutar concludes:

“The data mostly do not support the part of the critical narrative that suggests the spread of capitalist economic institutions was calamitous for living standards. Instead, they mostly corroborate a narrative, according to which living standards were poorer before the transition to capitalism and started improving afterwards”.

Before even considering causal empirics, correlational data is quite suggestive of what we already know about economic freedom and standards of living. Ample correlational data is available courtesy of the Fraser Institute’s Economic Freedom of the World report.

In countries with greater economic freedom, citizens enjoy substantially higher incomes. Those in the freest 25% of countries earn, on average, about 6.2 times as much as those in the least free, as shown in figure 1.

Figure 1. Economic Freedom, GDP per Person and Country Size

The level of income earned by the poorest 10% of the population is much higher in countries with greater economic freedom. The bottom 10% income threshold is 7.8 times higher in the freest quartile than it is in the least free, as shown in figure 2.

Figure 2. Economic Freedom and Bottom 10% Income Threshold

Figure 3 shows that the rate of poverty in the least-free quartile is about 25 times greater than it is in the freest.

Figure 3. Economic Freedom and Poverty

As figure 4 shows, people in the freest quartile live about 17 years longer than those in the least-free quartile.

Figure 4. Economic Freedom and Life Expectancy

In the least-free countries, infants die at nearly 10 times the rate they do in the freest countries, as shown in figure 5.

Figure 5. Economic Freedom and Infant Mortality

Economic freedom is positively correlated with personal freedoms, as shown in figure 6.

Figure 6. Economic Freedom and Personal Freedom

People in economically free societies are more satisfied with their lives, as shown in figure 7.

Figure 7. Economic Freedom and Life Satisfaction

While the newly published BMJ study presents no causal evidence for its conclusion that capitalist reforms hurt living standards, there is extensive empirical, causal evidence that, in most cases, the opposite is true. Vincent Geloso, a professor of economics at George Mason University, created an extensive thread of empirical studies that demonstrate how economic liberalization has boosted growth and enhanced living standards.

A 2013 article published in the Review of Economics and Statistics, for example, used a synthetic controls method (SCM) to demonstrate that most episodes of liberalization enhanced economic growth. Further improvements on the SCM, combined with replications of this 2013 article, corroborate the main findings.

A number of strong papers employ matching methods rather than the SCM. These approaches are appealing because they focus on discrete, “large” increases in economic freedom (using data available since 1970), and they pair economically reforming countries with the closest nonliberalizing counterparts.

In doing so, matching-method papers also mitigate concerns about heterogeneous treatment effects, which can bias some difference-in-differences (DiD) estimates. For example, a 2021 article in the Journal of Comparative Economics found that countries with sustained economic freedoms were 16% richer 10 years later, as shown in figure 8.

Figure 8. Real Per Capita Income Before and After Economic Reform

However, this estimate may itself be conservative due to the tendency of dictatorial regimes to overstate their GDP level. Recent work by Vincent Geloso and coauthors uses nighttime light data to construct alternative measures of economic activity, following the approach of economist Luis R. Martinez, and reassesses the magnitude of these effects. They find that the true impact of economic freedom on income levels is likely between 1.1 and 1.62 times larger than conventional estimates.

Applied to the earlier figure (16%), this research implies gains closer to roughly 18% to 25% after 10 years. In other words, measurement limitations in standard GDP data may lead the literature to systematically understate the benefits of liberalization.

Other empirical studies have estimated how economic freedom affects the probability that countries move between different income-growth “states,” such as escaping a poverty trap or falling back into one. Using a panel logit model, economists Donatella Saccone and Matteo Migheli find that greater economic freedom generally increases the likelihood of transitioning to higher-income, higher-growth clusters and reduces the risk of remaining trapped in low-income, low-growth states.

Importantly, the economic benefits of liberalization depend on sequencing and context. If liberalizing reforms are applied unevenly, then the results are likely to be disappointing. Imagine a regime that reduces tariffs but fails to establish private property rights.

One 2024 empirical exploration of 49 economic freedom-enhancing reforms found that 10 out of 49 cases experienced low or negative growth post-reform. The authors found that low levels of individualism, low levels of generalized trust and less focus on sound monetary policy were among the key underlying issues driving these disappointing results.

Another key issue is credibility. Autocrats who liberalize the economy can just as easily reverse course, often on short notice. Businesses, workers and investors understand this risk, which dampens their response to reforms. If property rights, market access or regulatory conditions can be withdrawn at any moment, the incentives to invest, expand or take risks remain limited.

Recent evidence supports this view. A study in the Journal of Comparative Economics finds that economic liberalization delivers much stronger results when it is paired with political liberalization. In cases where markets are opened without accompanying democratization, the gains tend to be modest. By contrast, when both economic and political reforms occur together, the benefits are substantially larger.

This finding suggests that disappointing reform episodes are not evidence against liberalization itself, but rather reflect deeper political economy constraints, especially issues of credibility, institutional quality and reform design.

Another possible critique that opponents of market liberalization make is that these reforms weaken institutions and government accountability. Using a doubly-robust DiD approach, recent empirical work finds that capitalistic reforms improve government accountability and imposes greater checks on government power, as shown in figure 9 below.

Figure 9. Response of Government Accountability to Economic Liberalization

Across countries, methods and decades, societies that protect property rights, allow markets to function and limit arbitrary state power are consistently richer, healthier and more upwardly mobile. Economic freedom is not a driver of poverty; it is one of the most reliable paths out of it.

What critics like Hickel mistake for failure is often something else entirely. When liberalization “fails,” it is rarely because markets were allowed to work; it is because they were not. Reforms that lack credibility, are poorly sequenced or are implemented under institutions that cannot sustain them will naturally produce weak results. But that is a critique of political constraints, not of economic freedom itself.

Smith calls the discovery of America one of the greatest events in human history. He also argues that the system built on that discovery—the American colonies—was a mistake.

How can both be true?

The answer lies in a distinction Smith draws—implicitly but consistently—between the expansion of markets and the political control of them. The former generates widespread gains. The latter distorts, redirects, and often destroys them.

Why Colonize in the First Place?

Smith is clear that nations do not establish colonies without expecting some return. He groups those expected returns into two categories: common advantages and particular advantages.

The common advantages are those that any extension of dominion is supposed to provide—greater military strength and additional public revenue. But in the case of European colonies, Smith argues that these advantages largely fail to materialize. The colonies of Europe, he notes, have “never yet furnished any military force for the defence of the mother country,” and in many cases have required defense themselves. Rather than strengthening the state, they divert military resources away from it.

The fiscal case is no stronger. With few exceptions, colonial revenues did not cover the costs of administration and protection, particularly in times of war. What was expected to be a source of strength was instead a source of expense. Noting England’s colonies specifically:

The taxes which have been levied upon [them]… have seldom been equal to the expence laid out upon them in time of peace, and never sufficient to defray that which they occasioned in time of war. Such colonies, therefore, have been a source of expence and not of revenue to their respective mother countries.

If colonies do not provide these common advantages, then their justification must rest on what Smith calls their particular advantages—namely, the exclusive trade secured by the mother country.

But this justification rests on a deeper mistake. It assumes that the gains from trade can be contained—that they can be captured and held within political boundaries.

As he puts it:

After all the unjust attempts, therefore, of every country in Europe to engross to itself the whole advantage of the trade of its own colonies, no country has yet been able to engross itself anything but the expence of supporting in time of peace and of defending in time of war the oppressive authority which it assumes over them. The inconveniencies resulting from the possession of its colonies, every country has engrossed to itself completely. The advantages resulting from their trade it has been obliged to share with many other countries.

Nations can monopolize political control. They cannot monopolize the economic gains from expanded trade. Those gains diffuse outward through exchange, while the costs of maintaining empire remain concentrated at home.

This failure is not accidental. It reflects something deeper about how the gains from trade operate: they are not confined to the parties who initiate exchange, and they cannot be contained within political boundaries.

You Don’t Need to Trade With a Country to Benefit From It

It is easy to see that when two countries trade, both benefit. It is less obvious that the discovery of a new market benefits countries that trade with it indirectly. Smith goes further still: countries can benefit from a new market even if they neither export to it nor import from it at all.

He illustrates this with Hungary and Poland—countries which may never have “sent a single commodity of their own produce to America.” And yet, he insists, their industry is nonetheless improved.

Here is Smith’s key passage:

Those commodities of America are new values, new equivalents, introduced into Hungary and Poland to be exchanged there for the surplus produce of those countries. By being carried thither they create a new and more extensive market for that surplus produce. They raise its value, and thereby contribute to encourage its increase. Though no part of it may ever be carried to America, it may be carried to other countries which purchase it with a part of their share of the surplus produce of America.

What Smith is describing is not bilateral trade, but a system of circulation. American goods enter the broader commercial system and expand the purchasing power of those who first receive them. That increase in purchasing power then spreads outward, enlarging markets well beyond the original point of exchange.

The same mechanism that allows Hungary to benefit without trading directly with America is what prevents Britain from capturing those gains.

Smith then pushes the point even further:

[The discovery and colonization of America] may even have contributed to increase the enjoyments, and to augment the industry of countries which not only never sent any commodities to America, but never received any from it. (emphasis added)

Once new goods enter circulation, their effects are not confined to where they first arrive. They expand the total mass of commodities available, lowering prices, raising real incomes, and encouraging further production—even in countries with no direct connection to the original trade.

This is why the attempt to monopolize those gains fails. Trade expands the size of the market itself—and while that expansion can be limited, it cannot be contained.

Why Colonies Persist

If Smith is correct that colonies impose net costs on the mother country, then their persistence requires explanation.

On economic grounds, the conclusion is unambiguous. “Under the present system of management,” he writes, “Great Britain derives nothing but loss from the dominion which she assumes over her colonies.” The case for separation, in purely economic terms, is straightforward.

And yet, Smith is equally clear that such a policy will not be adopted:

To propose that Great Britain should voluntarily give up all authority over her colonies… would be to propose such a measure as never was, and never will be adopted, by any nation in the world. No nation ever voluntarily gave up the dominion of any province… Such sacrifices… are always mortifying to the pride of every nation…

The obstacle is not economic—it is political. To abandon the colonies would require admitting that the system itself had been mistaken, at a cost to national prestige and to those who benefit from its administration.

Smith emphasizes this point:

…they are always contrary to the private interest of the governing part of it, who would thereby be deprived of the disposal of many places of trust and profit… of many opportunities of acquiring wealth and distinction…

The persistence of colonial rule is therefore not a puzzle once we shift from national interest to political economy. What is costly to the nation as a whole may still be beneficial to those who govern it.

That logic is not confined to colonial systems. Policies today are often sustained not because they generate broad economic benefits, but because they serve concentrated interests or avoid politically costly admissions of error.

Smith’s critique is therefore not just about empire, but about the persistence of policies whose economic case has collapsed but whose political foundations remain intact.

This article summarizes the main findings from a recent policy brief in which I examine the economic case for tax simplification in greater detail.

Compliance alone imposes a significant burden. In 2024, Americans spent an estimated 7.9 billion hours complying with federal tax filing requirements. This translates into roughly $413 billion in lost productivity, alongside an additional $133 billion in out-of-pocket compliance costs. In total, tax compliance costs approach $546 billion annually, or nearly 2% of GDP. These are real resources diverted away from productive activity.

But the costs of the tax code are not limited to compliance. Over time, the tax base has been narrowed through the expansion of deductions, credits, exemptions and exclusions, often together referred to as “tax expenditures.” These provisions function as implicit subsidies, favoring particular activities or industries. In doing so, they distort relative prices and redirect capital toward less productive uses.

Tax reform should not be evaluated solely in terms of statutory rates. The broader design of the tax base, and the incentives it creates, matter at least as much as do tax rates for economic performance.

Why Base Broadening Matters for Growth

Tax systems that broaden the base and reduce distortions tend to improve long-run economic outcomes.

Simulation-based studies of U.S. tax reform provide a useful starting point. Reforms that eliminate most deductions and credits while maintaining revenue neutrality are typically associated with sizable increases in output. These gains arise not from lower tax rates per se, but from improved incentives for work, saving and investment. When the tax system no longer favors specific activities, resources are reallocated toward their most productive uses.

Empirical evidence from real-world reforms reinforces this conclusion, with studies finding that such reforms raised annual growth rates during transition periods, primarily through increased investment.

Taken together, the literature suggests that reducing tax-induced distortions, especially those embedded in the tax base, can produce meaningful gains in economic performance.

The Hidden Costs of Complexity

Tax complexity imposes additional costs that are often less visible but equally important:

1. Complexity weakens perceptions of fairness. Survey evidence shows that many taxpayers view a complex tax system as inherently less equitable. When individuals believe that similarly situated taxpayers are treated differently, voluntary compliance declines, and incentives for evasion increase.

2. Complexity raises the cost of entrepreneurship. Navigating a complicated tax code requires time, expertise and resources that disproportionately affect small firms and new entrants. Empirical studies show that more complex tax systems are associated with lower rates of business formation and reduced economic dynamism. Even modest reductions in administrative burden can significantly increase the rate of new firm creation.

3. Complexity contributes to enforcement challenges. The U.S. tax gap (the difference between taxes owed and taxes paid) now approaches $700 billion annually. While multiple factors contribute to this gap, complexity makes accurate reporting more difficult and enforcement more costly.

In short, complexity is not merely an administrative inconvenience. It is a drag on economic efficiency, entrepreneurship and growth.

Investment, Capital and Growth

The most important channel through which tax simplification affects growth is investment.

Under the current system, savings and investment are often taxed multiple times. Income is taxed when earned and again when returns are realized, and while recent reforms have improved cost recovery, the tax code still falls short of full expensing in many cases. Together, these features raise the “user cost of capital”—the effective price firms face when investing.

A large body of empirical research finds that investment is highly responsive to changes in this user cost. Once methodological issues are addressed, estimates of the long-run elasticity of investment generally fall between −1 and −2. This implies that a 10% reduction in the cost of capital can increase investment by roughly 10-20%.

A simplified, consumption-based tax system, such as the Hall–Rabushka flat tax, would eliminate many distortions. By allowing full expensing and removing taxes on the normal return to capital, such a system would reduce the user cost of capital substantially.

The implications for growth are significant. A reduction in the user cost of capital on the order of 10% to 20% could increase the long-run capital stock by roughly 15% to 40%. Using standard production relationships, this translates into a GDP increase of approximately 8% over time.

Policy Implications

The case for tax simplification is often framed in terms of fairness, transparency or administrative ease. These are important considerations, but they understate the broader economic stakes.

A complex, distortionary tax system discourages investment, misallocates capital and suppresses entrepreneurship. A simpler system, one that broadens the base and treats saving and investment neutrally, can improve incentives across the economy.

Importantly, recent U.S. tax policy has already moved partially in this direction. The Tax Cuts and Jobs Act introduced full expensing for certain investments, reducing the tax penalty on capital formation. Evidence from the period when the statute was enacted shows that investment responded strongly to these changes, consistent with the broader empirical literature.

The One Big Beautiful Bill Act made these reforms permanent; however, they remain incomplete and uneven. A comprehensive approach to tax simplification would extend these principles across the tax code, eliminating many existing distortions.

Tax simplification should be understood not as a narrow administrative reform, but as a pro-growth policy. The current U.S. tax code imposes large compliance costs, embeds extensive distortions and raises the cost of investment. These features reduce economic efficiency and limit long-run growth.

By contrast, a simpler tax system, one that broadens the base and removes penalties on saving and investment, can generate substantial increases in capital formation and output. Even under conservative assumptions, the potential gains are economically meaningful.

A tax system that interferes less with economic decision-making is not only simpler, but more conducive to growth.

In 2023 alone, 6.7 million Americans packed up and moved across state lines. Among them: roughly 700,000 high earners ($200,000+), taking their income, their tax dollars, and their economic gravity with them.

So where are they going? And who’s losing them?

The Winners

Florida and Texas are running away with this. Florida netted 113,494 new residents, including 50,485 high earners, bringing a staggering $20.7 billion in income into the state. To put that in perspective, more than $17 billion of that came from high earners alone.

Texas came in close behind with 111,079 net new residents and $5.3 billion in net new income. But here’s an interesting difference: Texas attracted far fewer high earners proportionally (15,470 vs. Florida’s 50,485). Texas is growing fast, but its migration story is broader and more working- and middle-class than Florida’s.

The Carolinas are quietly becoming migration powerhouses. North Carolina gained 69,001 residents and $3.9 billion in income; South Carolina gained 58,610 and $4.1 billion. Tennessee added 42,401 residents and $2.8 billion. Georgia rounded out the list with 34,800 net new residents, but total income of negative $35 million (net positive $1.9 billion from high earners). This suggests the overall mix of Georgia movers skews toward lower-income transplants, even as it attracts a meaningful number of wealthy arrivals.

Arizona rounded out the winners at 26,215 residents and $2.7 billion.

The Losers

California lost 205,788 residents, the largest net outflow in the country, including 37,777 high earners. They took $13 billion in income with them, $7.6 billion of that from the high-earner cohort alone. That’s not just people leaving, but a significant narrowing of the tax base.

New York wasn’t far behind, losing 161,963 residents and $10.6 billion in income. Between California and New York, that’s nearly 368,000 people and $23.6 billion in annual income gone in a single year.

Illinois (-54,881 residents, -$6.2 billion), New Jersey (-31,456, -$2.8 billion), Massachusetts (-29,870, -$4.2 billion), and Maryland (-20,578, -$1.9 billion) round out the losers. Notably, every single one of these states has high income taxes, and state policymakers that continue to squeeze their states highest earners for more and more revenues.

What This Actually Means

State tax revenues are heavily tied to the income taxes paid by high earners. When a state loses nearly 38,000 high-income filers in a single year, as California did, it doesn’t just lose those people’s tax contributions. It loses the downstream economic activity they generate: the restaurants they eat at, the employees they hire, the properties they buy.

Meanwhile, states like Florida, which has no income tax, are compounding their advantages. More high earners move in, raising property values and consumer spending, which funds government through sales and property taxes. It’s a virtuous cycle for the state, and a vicious one for those bleeding residents.

One data point that deserves a double-take: New Jersey lost $2.8 billion from all movers but $2.9 billion from high earners alone. That’s not a typo. It means that lower-income movers into New Jersey partially offset the high-earner losses, but the state is still in the red on net income. The people leaving are disproportionately the ones who pay the most in taxes.

The Bottom Line

If you’re looking for a single sentence summary: high-tax, high-cost blue states are exporting their wealthiest residents to low-tax Sunbelt states, and the IRS data proves it at scale.

This trend predates COVID, but the pandemic supercharged it, and remote work broke the geographic tether that kept high earners in expensive metros. Three years later, the data confirms those movements weren’t temporary.

The notion that we are in a uniquely broken system, and unprecedented times can only be the result of economic and historical ignorance. Modern America is almost obviously more stable and better off than it was during the 1960s and 1970s. On almost any given metric (crime, economic, health, etc.), Americans are living better today than they were 50 years ago.

Political violence was rampant in the 60s and 70s. The 1960s was a decade for assassinations: President John F. Kennedy, Lee Harvey Oswald, Malcolm X, Martin Luther King Jr., and Robert Kennedy all lost their lives to assassinations. The 1970s was a decade for bombings. There were approximately 2,500 bombings from 1971 to 1972, and there were 1,471 recorded terrorist attacks during the decade, more than doubling the number of attacks in any decade that followed.

The 1970s were a time of high and rising crime in general. From 1970-1979, there were on average per 100,000 people: 452 violent crimes, 4297 property crimes, 9 murders, 26 forcible rapes, 196 robberies, 221 aggravated assaults, 1,340 burglaries, 2,498 larceny/thefts, 495 vehicle thefts. From November 2024 to October 2025, there were on average per 100,000 people: 334 violent crimes, 1,592 property crimes, 4 murders, 37 rapes, 52 robberies, 241 aggravated assaults, 203 burglaries, 1,180 larceny/thefts, and 209 motor vehicle thefts, and all these crime rates have gone down since the previous year.

The economy is also far stronger than it was 50 years ago. Real GDP per capita has jumped from $28,061 to $69,756 from 1975-2025. From 1970-1981, the average annual inflation rate was 7.3%, with rates peaking at 12.4% in 1980. The average annual inflation rate was just 3% from 2015-2025, peaking at 6.2% in 2022. People have also been getting richer in real terms: real median personal income from 1974-2024 went from $28,720 to $45,140 and real median household income went from $60,420 to $83,730 from 1984 to 2024.

People also live longer now. In 1970, average life expectancy was 70.8 years, in 2025 it was 79.4 years. More safety features and medical advances have played their part in extending lifespans. GLP-1s have started to reverse the trend of rising obesity. From 1970 to 2023, traffic fatalities dropped from 52,627 to 40,901, and 4.1 per 100 million vehicle miles driven to 1.26.

Americans have increasingly had the money and opportunity to see the world. Estimates in 1968 suggested that only 6% of Americans had traveled internationally; now over 70% have been to another country. The share of Americans who have valid passports has jumped from 3% in 1989 to 51% in 2024.

Concerns regarding housing proliferate amongst the discontent, but issues with housing have more to do with demographic shifts and regulation than systemic market failure. Other than surveys done by the National Association of Realtors, measures of the median age of first-time home buyers have stayed constantly in the early 30s throughout the past decade. Housing costs have grown less than average wages since 2000. Median home prices grew faster than median household incomes from 1978-2000, but actually grew slower than median household incomes when controlling for home quality.

Home ownership rates for young adults (25-34) have declined since the 70s but have held constant when controlling for relationship status. It is easier for two people to buy a home together than to buy one as an individual. The decline in young people getting married is a significant contributor to the difficulty young people have with buying a home. However, there is still room for improvement in making the home-buying process more accessible. Financial regulations increase concentration in the mortgage lending market and zoning restrictions limit the supply of new housing. Removing these regulations would make housing more accessible.

All these points do not prove that America is perfect or that there is nothing to be worried about. They do, however, mean that we should not forget to count our blessings. America is far safer and more prosperous than it was 50 years ago. One can still push for making things better without making the false claim that everything has gone downhill in America over the past 50 years.

It is politically advantageous for politicians and media personalities to play up the bad aspects of our modern world. Outrage at the state of the world is frequently used as the pretext for the government to step in to provide a “solution” and attracts attention online. Comparing the state of the world to how things were just 50 years ago cut against that outrage. Americans have many things to be thankful for and should take prophecies of doom with a grain of salt.

Caleb Petitt is a research associate at the Independent Institute in Oakland, Calif. @CalebDPetitt