Stock

Ryan Bourne

New York Times columnist Paul Krugman says we shouldn’t obsess about the federal debt. If his is the best case for not worrying, perhaps we really should panic.

As John Arnold noted, Krugman’s argument is essentially three‐​pronged. First, he notes that other countries have had similarly high or even higher debt levels before, including Britain after World War II. Second, he thinks stabilizing debt relative to gross domestic product (GDP) is economically simple—it “just” requires reducing the federal deficit by 2.1 percent of GDP every year (that’s about $600 billion right now). Third, he concludes that Republicans who express concern about the debt usually advocate tax cuts that worsen the debt outlook, so why should anyone else bother?

As Arnold noted dryly, “well, that alleviates my concerns”!

The case study of how Britain reduced its extraordinarily high post–World War II public debt (at 270 percent of GDP, falling to less than 50 percent by the mid‐​1970s) alone shows why these points are of scant comfort. The federal government debt held by the public, at 99 percent of GDP today, is certainly lower than Britain’s was. Even on unchanged policies it is projected to rise to “only” 166 percent of GDP by 2054. Yet none of the conditions that saw Britain reduce its massive debt burden quickly and steadily apply to us now:

British public spending plummeted from 62.4 percent of GDP in 1944–45 to 39.6 percent of GDP in 1954–55 as the country demilitarized and spending then remained relatively stable compared to GDP for two decades afterwards; in contrast, US spending is projected to rise from 23.1 percent of GDP to 24.1 percent by 2034 and then further to 27.3 percent by 2054.
Britain’s politicians committed to balancing budgets, running large primary budget surpluses (i.e., surpluses excluding interest payments) for a near quarter of a century after the war ended that led to modest overall deficits (less than 1 percent of GDP) throughout that period; in contrast, the United States is running a 2.5 percent of GDP primary deficit today and has an extremely large and growing overall deficit (already above 5 percent of GDP) that will grow as debt costs rise with all this new borrowing.
Britain experienced reasonable economic growth in the three decades after World War II, with real GDP growth averaging 2.3 percent per year; in contrast, the Congressional Budget Office projects the headwinds of an aging population and slow trend productivity growth will see sustainable real GDP growth fall to 1.8 percent annually for the United States within a decade.
Britain saw some major bursts of damaging inflation immediately after World War II and then again in the 1970s that eroded the real value of government. In fact, inflation averaged 6.2 percent per year for the three decades after the war ended; in contrast, while the recent inflation alone may have eased the federal debt burden somewhat, the key long‐​term drivers of debt—Social Security and Medicare—are inflation‐​protected, meaning unexpected inflation won’t much help to ameliorate our debt path.
Britain used financial repression policies to help force government interest rates below inflation, meaning the government benefited from negative real interest rates for 24 of the 30 years after World War II; in contrast, American financial markets are more sophisticated, and although it’s unclear where borrowing costs will end up, real interest rates are positive right now.

In essence, we have been borrowing as if we’ve faced an existential war but have no prospect of the equivalent of demilitarization, or any of the other favorable conditions that allowed Britain to reduce its debt burden sustainably. In fact, we still face the risks of future unforeseen wars, pandemics, or other sharp economic downturns that could blow up debt further and precipitate the implicit or explicit debt crisis that Krugman dismisses.

All these fundamentals suggest that, to avoid ever‐​escalating debt (which nobody thinks is sustainable), we are going to have to see substantive spending cuts or tax revenue rises. Krugman is relaxed that stabilizing the debt path through “just” $600 billion in deficit reduction today is achievable, but his own piece then argues against that premise. He believes that Republicans won’t raise taxes, but just this week even relative deficit hawk Rep. Ro Khanna (D‑CA) said he wouldn’t countenance any cuts to social spending to reduce deficits. In fact, House Democrats regularly introduce bills to increase old‐​age entitlements.

So how, exactly, will the debt be defused? That seems to me a challenge worth obsessing about.

For more on the federal debt challenge, read “A Case for Federal Deficit Reduction.”

Vanessa Brown Calder

In recent months, Alabama’s Supreme Court ruling on in vitro fertilization (IVF), the subsequent passage of a bill protecting fertility treatment in the state, and proposed federal IVF legislation have all generated new interest in a topic area once taken for granted by policymakers and policy analysts. Suddenly, IVF critics feel empowered to voice their skepticism about a popular medical process used to expand US families.

While certain Republicans have distanced themselves from the issue, some conservative and pro‐​life groups have leaned into the IVF opposition. Last week, the Clare Booth Luce Center for Conservative Women hosted an event on the bioethics of IVF, where Heritage Foundation’s Emma Waters presented her critical views on the topic.

The event highlighted many themes and claims in Waters’ recent written work. Unfortunately, Waters’ work is heavy on assertions and light on credible facts and nuance.

Like many critics, one of Waters’ primary concerns is that IVF results in embryo loss. Perhaps in an attempt to exaggerate the extent of embryo loss, Waters claims that a single cycle of IVF creates “10 to 15 embryos on average,” resulting in 4.1 million embryos created per year by her estimate.

Yet these figures are unrealistic, given that the reported average number of eggs retrieved is between 8 and 15—a 2011 study found the median number was 9—and those familiar with the process know the number of eggs retrieved is not equal to the number of embryos produced.

Instead, the potential number of viable embryos produced by a given number of eggs declines at each stage of development. First, because not all retrieved eggs will be mature and therefore suitable for fertilization, then because not all mature eggs will successfully fertilize, and finally because not all fertilized eggs will develop into blastocysts that can be transferred to the uterus (two‐​thirds of embryos’ development arrests). Given attrition rates at each stage, retrieving a median number of eggs would result in less than two viable embryos per cycle.

The imperfectness and inefficiency of conception under IVF makes it more similar to conventional reproduction than IVF critics care to admit. As the research study “Genetic Considerations in Recurrent Pregnancy Loss” states, “Human reproduction is remarkably inefficient; nearly 70% of human conceptions do not survive to live birth.”

In addition to exaggerating embryo loss, Waters also exaggerates the cost and risks associated with treatment. Waters claims that “on average, the complete cost of IVF is $72,000, as it can take multiple rounds to be successful.” Yet a US Department of Health and Human Services (HHS) fact sheet published this year states that one cycle of IVF is around $15,000 to $20,000, and an average of 2.5 treatment cycles are needed to become pregnant from IVF.

With that in mind, the average cost of successful IVF is between $37,500 and $50,000, without considering any options that reduce treatment cost, including outcome‐​based (money‐​back) programs that major clinics offer, IVF grants, discount programs, and the like. HHS likewise states that the average cost can “exceed $40,000.” (Note that in high‐​cost markets, like Northern California, average cost could be higher, but this is by no means widely generalizable.)

It is clear that fertility treatment is not cheap, but $40,000 or more is quite different from $72,000. Furthermore, in the context of other medical procedures ($442,500 for a kidney transplant), $40,000 no longer seems like such an outlier. Most people consider creating new human life a valuable pursuit, making high costs worth it for many would‐​be parents.

In addition to exaggerating figures, Waters characterizes IVF in frightening ways, including describing IVF as the “Wild West” of reproductive technology. Yet assisted reproductive technology is regulated at the state and federal levels and through professional societies like the American Society for Reproductive Medicine, Society for Assisted Reproductive Technology, College of American Pathologists, American Board of Obstetrics and Gynecology, and American Board of Urology.

Waters likewise suggests that preimplantation testing is used to “assess the embryo’s intellectual aptitude … how smart will the embryo be,” which sounds dystopian. But whether or not current technology is capable of making accurate predictions about intelligence, the concept is controversial, and neither widely available nor widely used.

Stunningly, Waters uses possible health outcomes like cancer, autism, and cleft palate as an apparent mark against IVF in her writing, even going so far as to quote a bioethicist who states “there is a world of difference between accepting the risk of a disabled child (where that risk is imposed upon us by nature) and ourselves imposing that risk in pursuit of our own purposes.”

Regardless of the cause, suggesting that human life is less worthy of pursuit due to the mere potential of health complications seems at odds with her organization’s ardent anti‐​abortion position and in tension with her opposition elsewhere to preimplantation testing (a tool that can reduce the incidence of life‐​altering or life‐​ending chromosomal or genetic abnormalities).

Finally, one of the least defensible arguments against IVF from a public policy perspective is that the practice doesn’t subscribe to a personal interpretation of biblical passages, as put forward in Water’s recent commentary. In a recent piece published on Heritage’s website, Waters used biblical passages, including the book of Genesis’ exhortation that “a man shall leave his father and his mother and hold fast to his wife, and they shall become one flesh” (Genesis 2:24) to suggest that IVF falls outside the biblical vision for procreation.

It seems unlikely that this passage or others seriously contemplate IVF’s morality, though that is largely beside the point. While biblical interpretation is one factor that may matter a great deal to an individual’s personal decision about whether or not to use IVF, an individual’s interpretation of Bible passages should not be a deciding factor in crafting policy in a pluralistic society where A) many individuals are not religious and B) those that are religious reasonably interpret biblical passages differently.

In short, exaggeration and misinformation undercut persuasion. Conservative critics should dispense with these strategies going forward, given the consequential effects on children, families, and infertile couples across the country.

Adam N. Michel

I recently testified before the Joint Economic Committee on why neutral, pro‐​growth tax cuts are a more effective domestic industrial strategy than targeted subsidies for specific firms, industries, or technologies.

You can read my full testimony here and watch it here. I was joined on the panel by my colleague Scott Lincicome, who highlighted similar themes.

Here, I will expand on one of the points I make in my longer testimony. The tax code’s bias against physical investments is set to worsen over the coming years and is a particular hindrance to the manufacturing sector.

Fixing the tax code’s treatment of investments should be Congress’ top priority.

Industrial Policies for Manufacturing

The US has recently shifted its economic policy toward a more active centrally planned industrial policy model. It has done this, in part, through the CHIPS and Science Act and the Inflation Reduction Act, which include trillions of dollars in direct subsidies that the Biden administration has paired with regulatory prohibitions and trade restrictions.

Ostensibly, the investment is intended to boost the American manufacturing base, whose demise is much exaggerated. But there is one way US manufacturing is struggling: sluggish productivity growth.

In a recent article, economist Joseph Politano explains that “recent industrial policy has thus far been unable to solve American manufacturing’s greatest fundamental problem—its complete lack of productivity growth.” Former director of the Congressional Budget Office, Douglas Holtz‐​Eakin, has noted a similar failure of recent federal efforts to jumpstart manufacturing productivity or manufacturing employment.

Focusing too narrowly on the manufacturing sector can distract policymakers from pursuing neutral, pro‐​growth policies that benefit all sectors of the economy. However, because the manufacturing sector tends to rely heavily on physical investments and research spending, it’s an instructive case study of how the tax code may be standing in the way of economic progress.

Tax policy is not the solution to all economic problems. Trade, immigration, labor composition, regulatory burdens, and institutions play critical roles in determining our economic fortunes, but getting tax policy right makes other economic headwinds easier to overcome.

The Tax Code Makes It Harder to Invest

The tax code is fundamentally biased against investment because it taxes investment returns more heavily than wages—through taxes on capital gains, dividends, business income, and estates. The tax code also effectively denies businesses the full value of deductions for expenditures on physical investments, such as equipment and buildings.

Businesses pay taxes on the income they earn, minus deductions for expenses—leaving taxable profits. If the tax code denies the full deduction for certain investments, it will artificially increase the cost of investing.

Assets that last longer (have longer asset lives, according to the IRS) face the biggest tax penalty. Figure 1 shows that at sustained 3 percent inflation, the present value of a $1 investment deduction falls quickly for longer‐​lasting assets.

For example, a business can only deduct about 88 percent of the real value of new equipment for vehicle manufacturing and assembly under the general depreciation system. Only 63 percent of investments in utility transmission infrastructure can be deducted.

The deduction value of a new manufacturing plant is eroded by as much as 60 percent over its 39‐​year write‐​off schedule. If a new semiconductor fabrication plant costs $1 billion and the new structure has to be depreciated over 39 years, Intel can only deduct roughly 1/39 (about $26 million) of what it paid to offset revenues in the first year. In 39 years, the final deduction of $26 million will be worth less than $3 million to the company. In the UK, the Adam Smith Institute has noted that a similar system creates a “Factory Tax” that largely falls on the investment‐​heavy manufacturing sector.

Generally, assets fall into one of eight classes, each with different asset lives between three and 39 years. The Tax Cuts and Jobs Act (TCJA) of 2017 temporarily allowed businesses to use the full deduction immediately (called full expensing) for short‐​lived assets (20 years or less). Beginning in 2023, this benefit phases out over five years, slowly making domestic investment more costly than it should be.

You can read more about expensing in my Cato Briefing Paper “Expensing and the Taxation of Capital Investment.”

Research Spending Plummets When Expensing for R&D Expired

Beginning in 2022, full expensing for research spending expired, requiring the costs to be amortized over five years (15 years for non‐​domestic expenditures). Research expenses include related wages for the researchers and their supervisors and other attributable costs, such as rent, utilities, and overhead. Five‐​year amortization effectively reduces the value of research deductions by about 13 percent (Figure 1), making research spending more costly after tax.

Figure 2 shows that the pre‐​COVID‐​19 average quarterly growth rate in real private R&D investment was 6.7 percent and 2021 R&D spending growth was even stronger. Following the loss of full expensing for R&D in the first quarter of 2022, R&D spending growth steadily declined, eventually turning negative. It averaged 2.3 percent through the first quarter of 2024—a 66 percent decline from the pre‐​covid average.

Research spending provides a stark example of how tax policy can significantly affect investment behavior. Policymakers should expect similar effects across the rest of the economy as expensing for the remaining short‐​lived assets continues to phase out over the coming years.

Bringing it Back to Manufacturing

Full expensing likely won’t solve the manufacturing sector’s slow productivity growth on its own, but it can make forward progress easier.

As we’ve seen over the last couple of years, targeted subsidies for particular types of investments tend to reshuffle existing resources toward politically favored sectors rather than expand the pool of total investment. There may be political returns, but the economic returns are usually disappointing.

Reforms like full expensing level the playing field between investments, expand investment opportunities, and lead to widely shared employment growth and wage gains.

In addition to making the TCJA full investment deductions permanent, Congress should also expand expensing to longer‐​lived structures, either by allowing the same immediate deduction or implementing a “neutral cost‐​recovery system,” which provides a similar economic benefit as expensing by allowing businesses to index their write‐​offs for inflation and time.

Despite increased investment in a few narrow sectors, targeted subsidies have not resulted in a broader manufacturing renaissance, nor should we expect them to. Instead of reaching for additional activist measures to boost politically popular industries, Congress should remove existing barriers to investment in the tax code and repeal existing targeted industrial policy subsidies.

Clark Packard

As Cato trade scholars have long lamented, the United States is much more protectionist than our relatively low average tariffs indicate, and one of the biggest offenders in this regard is a US “trade remedy” regime that imposes high duties on imports without regard to how they might harm American consumers, the broader US economy, or other strategic national interests. A recent case in the UK shows just how important such a safety valve could be.

On his Substack, Most Favoured Nation, British trade analyst Sam Lowe highlighted a recent decision by the United Kingdom’s Trade Remedies Authority (TRA) to recommend lifting existing trade remedies (antidumping and anti‐​subsidy duties) on imported Chinese electric bicycles (e‑bikes). The TRA based its decision (largely) on the grounds that the existing trade remedies don’t meet a cost‐​benefit analysis—that is, they hurt consumers more than they help UK e‑bike producers. For American trade watchers, such a decision is basically unheard of—but it’s worth emulating.

Here’s the backstory as laid out by the TRA’s Statement of Essential Facts in the antidumping and countervailing duties cases: As the British government was preparing to leave the European Union, the TRA analyzed whether existing EU trade remedies should be scrapped, adopted, or modified by the post‐​Brexit government. (The TRA has the authority to investigate existing as well as prospective trade remedies). As part of that process, the UK adopted the EU’s antidumping and countervailing duties (AD/​CVDs) on e‑bikes from China at the end of 2020. As Lowe notes, the current antidumping duties range from 10.3 percent to 70.1 percent and the countervailing duties range from 3.9 percent to 17.2 percent on covered e‑bikes. It was estimated that imports of Chinese e‑bikes into the EU fell by more than 90 percent after the imposition of the AD/​CVDs.

In late May of 2023, TRA began a probe into whether the e‑bike AD/​CVDs continue to make sense. In its recently published findings, the TRA noted that if the tariffs were lifted, it is likely the subsidization and dumping of e‑bikes would recur; that such subsidization and dumping would cause injury to UK e‑bike producers; but—most importantly—on net, “keeping [the tariffs] in place would not be in the economic interest of the UK” under the TRA’s ability to make determinations about the broader economic impact of trade remedies known as the Economic Interest Test (EIT).

The TRA determined that revoking the trade remedies would “benefit the UK economy by an average of £51m per year; save consumers an average of £260 per e‑bike; and result in an average of 31,000 more e‑bikes being bought per year in the UK.” In other words, the TRA determined that the consumer benefits and broader economic benefits of lifting the tariffs outweighed the gains for producers that would continue to accrue if the AD/​CVDs were kept in place. What a novel concept!

Following the TRA’s publication of the Statement of Essential Facts, other UK government agencies will have twenty‐​one days to weigh in on the proposed revocation of the e‑bike AD/​CVDs. If no other agencies oppose the TRA’s decision, the duties will be terminated by a final decision in September.

As Lowe notes, the case has wider implications for the UK, particularly with respect to Chinese electric vehicles (EVs). Could the TRA make a similar recommendation that AD/​CVDs on imported Chinese EVs fail a cost‐​benefit analysis under the EIT? Though there are substantial differences between e‑bikes and EVs, it’s possible.

Implications for US Trade Policy

Unlike the UK, the United States’ AD/CVD regime specifically prohibits consideration of the broader economic effects of imposing such trade remedies like the TRA’s EIT utilized in the case of the Chinese e‑bike. Instead, it is limited to questions of whether a product was subsidized and/​or dumped into the US market and whether such subsidization and/​or dumping injured a domestic industry (with no regard for other considerations). As a result, US trade remedies are often at odds with wider policy goals—and their costs frequently outweigh their benefits.

Two examples: The US continues to levy aggressive AD/​CVDs on imported solar products—stunting deployment by raising prices—even though politicians argue that climate change poses a grave risk to the United States and the world. Likewise, the Biden administration has recently argued the US needs to build more housing to help alleviate soaring costs, but it continues to maintain decades‐​long AD/​CVDs on Canadian softwood lumber imports, an essential component of residential housing. And it’s not just softwood lumber. A 2022 Cato paper from Alessandro Barattieri and Matteo Cacciatore found that the US levies numerous AD/CVD measures on a wide range of construction inputs with median duties ranging from 28 percent to 134 percent, all of which inflate the cost of residential building. If policymakers were truly serious about addressing climate change and housing affordability challenges, nixing these duties would be a priority. There are countless other examples.

More broadly, the US AD/CVD regime imposes significant costs on the economy. A 2020 working paper from Barattieri and Cacciatore found US AD/CVD actions between 1994 and 2015 largely targeted industrial inputs; depressed employment in downstream industries like those utilizing steel; and provided little long‐​term benefits for the jobs protected, such as in steel production.

Not only that, the use of trade remedies has also skyrocketed in recent years. As my Cato colleague Scott Lincicome has noted, there were about 300 AD/CVD orders in place as of 2016. Today, there are more than 675 orders on the books with 123 cases pending. In the next year or so, the US could have more than 800 AD/CVD duties on the books—a massive increase in a short period. See Figure 1 for the composition of current AD/CVD orders, by commodity.

The US trade remedy regime is in desperate need of reform. Permitting a true cost‐​benefit analysis of potential and existing trade remedies—similar to the UK’s system—is an ideal place to start.

Jeffrey A. Singer

The Centers for Disease Control and Prevention recently reported that, during 2023, the middle and high school student population using e‑cigarettes during the past thirty days dropped to just 7 percent, and teen e‑cigarette use fell from 14 percent to 10 percent between 2022 and 2023. Yet, the Senate Judiciary Committee plans to conduct hearings tomorrow on “Combatting the Youth Vaping Epidemic by Enhancing Enforcement Against Illegal E‑Cigarettes.”

The hearing will take place just two days after the Justice Department and Food and Drug Administration announced they have created a multi‐​agency coalition of law enforcement organizations that includes the Bureau of Alcohol, Tobacco, Firearms, and Explosives, the US Marshals Service, the Federal Trade Commission, and the US Postal Service to go after the underground trade in vaping materials.

Both the legislative and executive branches are in a hurry to solve the teen vaping epidemic before the public realizes that there isn’t one. As Michelle Minton wrote for the Reason Foundation last December, “The youth vaping epidemic, declared by the Food and Drug Administration (FDA) in 2018, appears now to have been more of a teenage fad—one possibly partially fueled by media attention on the issue. But, while the vaping fad may be subsiding, the hysteria surrounding it continues unabated.”

Importantly, a 2021 Brown University study found:

Vaping is largely concentrated among non‐​smoking youth who would likely have smoked prior to the introduction of e‑cigarettes, and the introduction of e‑cigarettes has coincided with an acceleration in the decline in youth smoking rates. E‑cigarettes may be an important tool for population‐​level harm reduction, even considering their impact on youth.

In an interview, one of the study’s authors stated, “The decline in youth smoking really accelerated after the availability of e‑cigarettes.”

This may help explain why teen smoking rates have hit historic lows.

I see a rationale for a new multi‐​agency law enforcement coalition as a jobs program for prohibitionists. But I don’t see any rationale for the scheduled Senate Judiciary Committee hearings on a non‐​existent “epidemic.”

David Inserra

The First Amendment and a robust culture of free expression mean most Americans understand that our society addresses hard topics through robust speech and debate. The way we handle disagreements among Americans is not to have the government wade into highly subjective, highly political issues to declare one group correct and silence the others as dangerous or wrong. Instead, we allow the free speech of Americans and the press to openly discuss and persuade each other of the truth or best course of action.

Unfortunately, this cultural dedication to expression is under pressure from government efforts to fund counter “disinformation” projects that are often thinly veiled efforts to suppress the speech of Americans. In response, Rep. Thomas Massie (R‑KY) introduced HR 8519 to prohibit the funding of research of disinformation and online trust and security. The Massie bill recognizes such funding for what it is: government funding of censorship. 

The short bill would prevent the government from funding three things:

(1) Disinformation research grants. 

(2) Secure and Trustworthy Cyberspace grants. 

(3) Programs within the National Science Foundation’s Track F: Trust and Authenticity in Communications Systems.

Before we get into what each of these prohibitions are, first a little bit of background. The terms mis‑, dis‑, and mal‐​information are similar and often used somewhat interchangeably as different experts and groups do not have uniform definitions. But here are the main differences:

Disinformation is the purposeful spread of false or misleading information designed to cause harm. This is often thought of in the context of governments trying to harm other nations or societies—indeed the term comes from the Soviet term “dezinformatsiya.”

Misinformation is the spread of false or misleading information but without the clear intent to do harm.

Malinformation is factual information that is out of context and thus could mislead (notice the overlap of “misleading” speech in each term depending on the definition that is used).

Together, these terms are often referred to as “MDM.”

But as high‐​profile incidents over the past few years have shown, MDM efforts and claims often got things wrong. Take claims about the source of COVID-19, the efficacy of vaccines in preventing sickness and the spread of COVID-19, debates over masks or school closures, the Hunter Biden laptop, and others. Even if well‐​intentioned, many claims of misinformation are little more than an effort to invalidate certain political or ideological views on important social issues. 

Yes, some misinformation is clearly false, but the more obviously wrong it is, the easier it is to refute. Most of the prominent and impactful MDM circulating today is just material that may be viewed as misleading, incomplete, or out of context, but it isn’t objectively false. It’s the opinions, biases, and contexts that any American may choose to express that some find persuasive while others find misleading. It’s the opinion page of the Wall Street Journal that emphasizes certain details, facts, and context while the New York Times editors emphasize different pieces of information about the same story to reach a different perspective. Researching misleading or out‐​of‐​context arguments is about as objective as researching whether the Times or Journal is more correct or whether MSNBC or FOX is better news.

So broadly forbidding government disinformation research grants is a great start for this bill. My only constructive criticism is that I would broaden the term to include the full trio of MDM research and any effectively similar terms or euphemisms. A clever politico or bureaucrat could argue that while disinformation funding is prohibited by this bill, it didn’t prohibit misinformation or malinformation funding. Out of an abundance of clarity and to prevent definitional gymnastics, I would expand the terms used, but this is right on target.

The final two lines are specific sets of funding within the National Science Foundation. Track F has funded the creation of AI tools that tech companies could use to combat MDM. As might be expected with research on subjective and political issues, the projects funded under Track F advance ideological biases under the guise of advancing “trust and authenticity” online.

Take, for example, the Co‐​Insights project funded by NSF, which presented itself as the “world’s best system for matching social media posts to facts.” The types of “common misinformation narratives” that it was working to counter were: 

Fearmongering and anti‐​Black narratives;

Undermining trust in mainstream media;

Glorifying vigilantism; and

Weakening political participation.

Each of these issues is linked to incredibly divisive political speech and events in which Americans have very different First Amendment‐​protected views. There is no reason for the government to be funding tools that seek to counter American viewpoints. 

If you don’t see a problem with the government funding this sort of research, imagine this: Your most hated political opponents have just been elected president and to a majority in Congress (which they will at some point). They are now able to funnel your tax dollars to fund efforts that label your views as misleading and harmful misinformation. For example, imagine the government funding research and tools to counter this alternative set of misinformation narratives:

Climate change catastrophism;

Anti‐​white CRT narratives;

Glorifying violent riots; and

Undermining trust in the independence of the Supreme Court.

No matter who the target is, this is censorial and wrong. 

The Secure and Trustworthy Cyberspace grants similarly include some inappropriate government funding of MDM research. While it also has other more technical and cybersecurity‐​focused grants, this is also a field where the private sector is already very active, so it is not clear we need to fund such projects generally and certainly not those that are working on misinformation. 

We also need to be vigilant for other ways that the government is funding the suppression of American speech. The State Department’s Global Engagement Center (GEC) funded the British Global Disinformation Index (GDI) that blacklisted American news organizations. The GDI created various lists of organizations that it considered spreaders of harmful misinformation and sold these lists to help big advertisers avoid these dangerous spreaders of misinformation. This resulted in any organization that didn’t match the GDI’s politics losing a significant amount of advertising revenue.

The GDI targeted mostly libertarian or conservative organizations, with the New York Post, the Daily Wire, Real Clear Politics, and Reason being listed as the highest‐​risk spreaders of misinformation. This is already in the courts and last year’s National Defense Authorization Act took aim at the GDI and similar entities at least for DOD funding. Rep. Massie’s bill might consider adding a broader restriction on government contracts—not just grants—along the lines of the NDAA’s limitation.

Many Americans feel overwhelmed by misinformation and long for help in determining what is true and false. Businesses, non‐​profits, and other private actors have dedicated significant time and money to stopping it, each with their own views on what should or should not be considered misinformation. Ultimately, it is up to each individual to be a good consumer of information and to critically analyze what is presented to them. 

The government cannot fairly adjudicate misinformation, nor can it fund research into such a subjective topic without trampling on the viewpoints of Americans. Rep. Massie’s bill is right to get the government out of the business of funding misinformation research.

Travis Fisher

Energy producers will be subject to retroactive taxes in New York if the state assembly passes Senate Bill S2129A, known as the “Climate Change Superfund Act.” The superfund legislation seeks to impose a retroactive tax on energy companies that have emitted greenhouse gases (GHGs) and operated within the state over the last seventy years.

If passed, the new law will impose $75 billion in repayment fees for “historical polluters,” who lawmakers assert are primarily responsible for climate change damages within the state. The state will “assign liability to and require compensation from companies commensurate with their emissions” over the last “70 years or more.” The bill would establish a standard of strict liability, stating that “companies are required to pay into the fund because the use of their products caused the pollution. No finding of wrongdoing is required.”

New York is not alone in this effort. Superfunds built on retroactive taxes on GHG emissions are becoming increasingly popular. Vermont recently enacted similar legislation, S.259 (Act 122), titled the “Climate Superfund Act,” in which the state also retroactively taxes energy producers for historic emissions. Similar bills have also been introduced in Maryland and Massachusetts.

Climate superfund legislation seems to have one purpose: to raise revenue by taxing a politically unpopular industry. Under the New York law, fossil fuel‐​producing energy companies would be taxed billions of dollars retroactively for engaging in legal and necessary behavior. For example, the seventy‐​year retroactive tax would conceivably apply to any company—going back to 1954—that used fossil fuels to generate electricity or produced fuel for New York drivers.

The typical “economic efficiency” arguments for taxing an externality go out the window with the New York and Vermont approach, for at least two reasons. First, the goal of a blackboard or textbook approach to a carbon tax is to internalize the GHG externality. To apply such a tax accurately, the government would need to calculate the social cost of carbon (SCC).

Unfortunately, estimating the SCC is methodologically complex and open to wide ranges of estimates. As a result, the SCC is theoretically very useful but practically impossible to calculate with any reasonable degree of precision.

Second, the retroactive nature of these climate superfunds undermines the very incentives a textbook tax on externalities would promote. A carbon tax’s central feature is that it is intended to reduce externalities from current and future activity by changing incentives. However, by imposing retroactive taxes, the New York and Vermont legislation will not impact emitters’ future behavior in a way that mimics a textbook carbon tax or improves economic outcomes.

Arbitrary and retroactive taxes can, however, raise prices for consumers by increasing policy uncertainty, affecting firm profitability, and reducing investment (or causing investors to flee GHG‐​emitting industries in the state altogether). Residents in both New York and Vermont already pay over 30 percent more than the US average in residential electricity prices, and this legislation will not lower these costs to consumers.

Climate superfunds are not a serious attempt to solve environmental challenges but rather a way to raise government revenue while unfairly punishing an entire industry (one whose actions the New York legislation claims “have been unconscionable, closely reflecting the strategy of denial, deflection, and delay used by the tobacco industry”).

Fossil fuel companies enabled GHG emissions, of course, but they also empowered significant growth, mobility, and prosperity. The punitive nature of the policy is laid bare by the fact that neither New York nor Vermont used a generic SCC or an evidentiary proceeding to calculate precise damages.

Finally, establishing a standard in which “no finding of wrongdoing is required” to levy fines against historical actions that were (and still are) legally permitted sets a dangerous precedent for what governments can do, not only to businesses that have produced fossil fuels but also to individuals who have consumed them.

Cato research associate Joshua Loucks contributed to this post.

Jack Solowey and Jennifer Huddleston

California lawmakers’ heavy‐​handed attempt to regulate cutting‐​edge AI development (SB 1047) received appropriate attention and backlash from the pro‐​innovation policy community. That backlash shone an important spotlight on the challenges state legislatures pose to AI innovation nationally.

Indeed, 40 states have considered some form of AI legislation this year alone, threatening to create an unworkable multi‐​state patchwork. The initiatives range from laws targeting specific AI applications (such as AI’s use in music under Tennessee’s ELVIS Act) to regulatory regimes for AI broadly.

In addition to legislation like SB 1047 designed to tackle frontier AI risk, another category of relatively broad AI legislation worth paying close attention to seeks to regulate AI’s use in so‐​called “consequential decisions,” such as employment, health care, and financial determinations.

While the risks these consequential‐​decision acts seek to tackle are seemingly more mundane than the putative threats of AI‐​enabled mass destruction targeted by frontier model legislation, the risks the consequential‐​decision acts pose to AI development are by no means trivial. They threaten AI development by undermining naturally emerging business models, putting AI developers in jeopardy, and targeting technologies not just harms.

Consequential‐​decision acts are gaining traction in the states. On May 17, Democratic Governor Jared Polis signed into law Colorado’s consequential‐​decision bill, albeit under a bit of protest. With the Colorado bill advancing into law before even the EU’s AI Act, the typically innovation‐​embracing state has come out ahead in a race it shouldn’t want to win.

On May 21, the California State Assembly passed its variation on the theme in a bill covering “Automated decision tools” (AB 2930), which is now before the California State Senate.

Though their details vary, both the Colorado and California consequential‐​decision acts seek to combat the risks of AI decision tools perpetrating algorithmic discrimination (which is roughly defined to mean unlawful disparate treatment or impacts disfavoring people based on their membership in a protected class). Specifically, the acts seek to combat discrimination when AI tools are used for decisions that have material, legal, or similar effects on access to things like education, employment, housing, utilities, health care, legal services, and financial services.

The Colorado and California regulatory approach addresses potential discrimination through a suite of obligations placed to varying extents on AI decision tool developers and deployers (i.e., the organizations using the tools in interactions with consumers). The acts generally impose duties—differing in their particulars—to avoid algorithmic discrimination, perform risk assessments, notify individuals regarding the use of AI decision tools, provide consumers with rights to opt out of and/​or appeal automated decisions and implement AI governance programs designed to mitigate the risks of algorithmic discrimination.

The Colorado and California acts’ automated decision opt‐​out and appeal rights provide a window into the two regimes’ similarities, subtle differences, and ultimate problems. Whereas the California bill creates a new automated decision opt‐​out right on top of an existing one mentioned in the state’s consumer privacy law, the Colorado law refers to a similar right in the state’s own privacy law while also adding new rights to appeal certain automated decisions.

Notably, California’s data privacy regulator has also begun preliminary rulemaking activity for California’s existing automated decision opt‐​out right. This points to a broader conversation that must be had regarding the interactions between data privacy laws and AI. Existing privacy regulations may not be well‐​adapted to the AI era. For instance, such laws’ data minimization requirements and limits on the use of personal information could undermine attempts to combat bias through more diverse data sets.

As for the opt‐​out/​appeal rights in the automated‐​decision acts themselves, generally, both bills require some form of alternative decision process or human review when it’s requested by a consumer and is “technically feasible,” but Colorado would require the consumer to wait for an adverse decision, while California is less clear on timing.

There’s something superficially enticing to many about circumventing automated decisions, but creating a blanket right to do so is not without costs. Indeed, automation often will be precisely what provides the cost savings that allow a business to offer products or services at an attractive price. Mandated opt‐​out rights likely would result in certain products and services becoming more expensive or unavailable.

Colorado’s more limited appeal right is a better approach but ultimately would impose similar costs, just to a potentially lesser degree. Furthermore, the caveat in the acts that alternative processes and human review be “technically feasible” is unlikely to help businesses with the technical ability to provide alternatives but without the resources to do so cost‐​effectively.

Absent opt‐​out mandates, businesses still would be able to provide such rights in response to consumer demand, while the broader ecosystem could simultaneously provide a greater range of features and prices.

The opt‐​out mandates’ constraint on naturally emerging business models is one of the core issues with the Colorado and California proposals. The others are the legal jeopardy and compliance burdens imposed on AI developers, as well as regulatory approaches that target technologies instead of harms.

The Colorado and California consequential‐​decision acts both impose onerous compliance risks and obligations on AI developers. Specifically, the acts inappropriately require developers, not just deployers, to anticipate and mitigate the risks of algorithmic discrimination. (Absurdly, the California bill even obligates AI developers to give legal advice to deployers, requiring developers to provide a “description of the deployer’s responsibilities under” the act.)

One major problem with this general approach is that it’s difficult, if not impossible, for a developer to completely understand an AI system’s propensity for discrimination in a vacuum or to predict every possible way their tool may be used. Any discriminatory effect of an AI system likely would be a product of both the underlying model and the deployer’s use, including the real‐​world data the deployer feeds into the model at the inference stage, as well as the deployer’s ability to implement compensating controls addressing any disparate outputs.

One way the acts address this problem is by cabining some developer obligations to only those risks that are “reasonably foreseeable.” Nonetheless, the California bill undermines this limitation by imposing a general duty on developers to “not make available” an AI decision tool “that results in algorithmic discrimination.” While the Colorado law does a better job of limiting developer duties to only reasonably foreseeable risks, it nonetheless has unreasonable expectations regarding what developers will be able to predict and take responsibility for. The Colorado law mistakenly assumes developers will have greater knowledge of an AI tool’s “intended uses” than is likely to be the case and requires developers to notify law enforcement after discovering their tool’s use by a deployer is likely to have caused algorithmic discrimination.

Requiring developers to orient their compliance measures around predicted use cases risks limiting the types of productive ends to which their models may be applied, as novel use cases could increase compliance risk. Disincentivizing developers from allowing all but the most obvious intended uses would be a huge loss for the AI ecosystem, as some of the most creative applications of technologies typically are devised downstream from the tool’s creator. That’s why, for example, third‐​party apps exist for smartphones.

Perhaps the original sin of the consequential‐​decision acts is that they target AI used for, well, consequential decisions. Such decisions tend to be those related to sectors that already are heavily regulated, such as health care and finance. For example, the core risk addressed by these acts—discrimination based on protected class membership—already is illegal in credit decisions under federal law. Targeting the technology, as opposed to the harm, in the financial services context, for instance, is redundant at best and counterproductive at worst, as it adds yet another layer of compliance burden that could stymie AI tools’ potential to expand credit access to the historically underserved. In addition, this general approach often misassigns the blame for bad or negligent actors’ improper use of technology to the technology itself.

This shortsighted regulatory playbook—constraining business models, burdening developers with responsibility for downstream risks, and targeting technologies instead of harms—is being employed all too often at the state level. After all, SB 1047 is a notorious vehicle for all three, making open‐​source AI development a compliance risk by requiring developers to lock down their models against certain downstream modifications, as well as targeting technical sophistication, not merely specific threats.

The risk from this playbook is that the US will be made worse off as state‐​level frameworks become de facto national standards without the benefit of national input. This is not just the case for legislation out of large states like California, as laws with long‐​arm ambitions and cloud‐​based targets can, in practice, extend compliance burdens beyond state borders. Where that’s the case, conflicting obligations and subtle variations can raise the question of whether full‐​scale compliance is even possible.

Instead of seeking to be the first to regulate, states should consider working from an alternative playbook that prioritizes innovation, avoids counterproductive interventions, and targets harms, not technologies. In the meantime, we should fear the words, “I’m from the state government, and I’m here to help with AI risk,” even when it’s another state’s government saying them.

Jon Hoffman

According to a new report in the Wall Street Journal, the Biden administration is close to finalizing a proposed treaty with Saudi Arabia that would include a formal US security guarantee for the kingdom and assistance from the United States in facilitating a civilian nuclear program, if Riyadh normalizes relations with Israel.

The treaty, the “Strategic Alliance Agreement,” is supposedly modeled on Washington’s security pact with other partners, such as Japan and South Korea, and would therefore require a two‐​thirds majority vote in the Senate for approval. However, Saudi Arabia has insisted that to move forward Israel must end its military campaign in Gaza and rhetorically commit to a “pathway” toward a two‐​state solution with Palestinians, something Israeli Prime Minister Benjamin Netanyahu has repeatedly rejected.

Though its future is uncertain, this effort by Washington should be denounced for what it is: an extension of a failed US Middle East policy.

President Biden’s unceasing drive for this “mega‐​deal” is rooted in the same magical thinking that has repeatedly brought the United States to ruin in the Middle East. This deal would achieve nothing for the United States, binding our regional policies to a principal source of instability in the Middle East, while planting the seeds for future disorder.

Saudi Arabia is not an ally—no amount of concessions to the kingdom will change this. Failure to abandon this disastrous agenda risks formalizing Washington’s commitment to a cycle of turmoil that will continue to impact the region—and undermine US interests—for generations.

Jeffrey Miron

A recent Supreme Court ruling on civil forfeiture, while seemingly a setback for critics of the practice, nevertheless contains some good news. 

Civil forfeiture occurs when police seize property they suspect was involved in or obtained through crime. The owner must then prove the property’s innocence to reclaim it.

States must ordinarily provide notice and a hearing before seizing property. If the property in question could otherwise be removed, destroyed, or concealed before a hearing, however, police can seize it without these steps. 

Civil forfeiture started during the colonial years under British maritime law. If a ship owner violated custom laws, enforcement could seize ships and cargo even if the owner was difficult to capture.

Forfeiture has expanded dramatically as a weapon in the War on Drugs (similarly to the Prohibition era). One crucial development has been that law enforcement agencies can often keep the proceeds from auctioning seized property, which creates an incentive to seize more property and complicate the retrieval processes. This seems likely to distort balanced enforcement of laws and is inconsistent with the presumption of innocence.

In a recent case about civil forfeiture, the Supreme Court ruled against two women in Alabama:

Halima Culley bought a 2015 Nissan Altima for her son to use at college. He was pulled over by the police in 2019 and arrested when they found marijuana. They also seized Ms. Culley’s car. That same year, Lena Sutton lent her 2012 Chevrolet Sonic to a friend. He was stopped for speeding and arrested after the police found methamphetamine. Ms. Sutton’s car was also seized.

Both women eventually convinced the courts to return their cars, but their cases took over a year. Culley and Sutton filed suit in federal court, arguing that, in the interest of “timely resolution,” civil forfeiture should allow property owners to prove “innocent ownership” in a preliminary hearing, before cases moved forwards. 

The court ruled 6–3 that while due process requires timely resolution, it does not always require a separate preliminary hearing. This is disappointing for critics of civil forfeiture, since defining and enforcing “timeliness” is messy (although the same could be true of preliminary hearings). 

The good news is that five justices expressed serious reservations about current practices.

Justices Gorsuch and Thomas, although agreeing that preliminary hearings should not be required, emphasized the abusive practices of today’s civil forfeiture:

[T]his promise [the Fifth and Fourteenth Amendments] usually meant that a government seeking to deprive an individual of her property could do so only after a trial before a jury in which it (not the individual) bore the burden of proof. […] So how is it that, in civil forfeiture, the government may confiscate property first and provide process later?

Justice Sotomayor, joined by Justices Kagan and Jackson, criticized the perverse incentives for officers to hold on to seized property.

A crucial step in reducing the negatives of civil forfeiture is repeal of drug prohibition. Then, police would have far fewer occasions to claim that property was involved in a crime.

With or without legalization, civil forfeiture would also be less onerous and more consistent with due process if: 

The government required a criminal conviction before seizing assets.

The government had the burden of proof.

Forfeited property went to the federal treasury rather than the police, to remove corrupt incentives.

Here’s hoping a future case leads to such changes.

Lemoni Matsumoto, an undergraduate at the University of Chicago, contributed to this article.