Conversable Economist

Forbes magazine has just published its annual list of the 400 wealthiest Americans, who this year hold a total of $8 trillion in wealth. It’s a gobsmacking amount of money, but we also live in a world where trillions are not unheard of. Here are a few thoughts and perspectives.

The promised Social Security benefits under current law exceed the projected inflows of revenue by about $29 trillion over the next 75 years, according to the actuaries. The promised Medicare benefits under current law exceed the projected inflows of revenue by about $60 trillion over the next 75 years. So if the US government somehow had an extra $8 trillion to play with, it could address either about one-quarter of the Social Security underfunding or about one-eighth of Medicare underfunding–but not both. My first point here is just that while US wealth held by the wealthiest is enormous, it is not an unlimited amount; indeed, the existing promises already made by just these two programs are much larger.

However, there’s no question that US wealth has been growing fast than the economy as a whole. This figure shows the ratio of wealth-to-GDP in the US since about 1960.

From the 1960s up through the 1980s, the ratio of total US wealth to the GDP was roughly 4–that is, with some fluctuations, total assets rose in proportion with the size of the economy. But in the the 1990s, this changed. There was a burst of stock market wealth in the 1990s and a burst of real estate wealth in the early 2000s, and in 2007 on the eve of the Great Recession, wealth was 4.9 times the size of GDP. After the Great Recession, the wealth-to-GDP ratio fell back to the historical level around 4. But a few years later, driven particularly by rising stock prices (which have risen almost six-fold since 2010), wealth-to-GDP took off again, topping out at a multiple of 5.5 in late 2021 and 2022, then sinking a bit, and now apparently pushing to a new high.

This pattern affects how one thinks about the rising inequality of wealth. If the wealth-to-GDP ratio remained the same over time, but the top 1% or 0.1% had a larger share, then the wealth gains at the top would come at the expense of wealth at the middle and the bottom. But with a rising wealth-to-GDP ratio, it’s possible for absolute quantity of wealth at the middle and the bottom to remain much the same, even as the quantity of wealth rises at the top. One might object that both scenarios are unpleasant, and I wouldn’t argue the point too vehemently, except to say that the two scenarios are unpleasant in different ways and with different implications.

The Federal Reserve now runs a website with detailed breakdowns of wealth data, updated quarterly. This figure shows the share of wealth held by different groups. this is household data, so with 134 million or so households in the US (and an average of 2.5 people per household), the top 1 percent consists of 1.34 million households. That top 1% held 8.6% of total wealth in 1989, and how holds about 15% of total wealth. Meanwhile, the bottom 90% of the wealth distribution held 77.2% of total wealth in 1989, but now holds 68.3% of total wealth.

This figure shows the types of assets held by different groups: again, remember that the number of households represented by the first bars, the top 0.1%, is only one-ninth as large as the second bar, and both are much smaller than the bars that follow.

The next figure shows the same data, but instead of presenting it in dollars, it shows the share of assets held by each group. What’s interesting here, I think, is that those at the top of the wealth distributino are more likely to hold assets in the form of stocks and “unincorporated businesses.” hare of assets held by each group. Those further down the distibution hold a greater share of wealth in real estate.

There are a number of proposals out there for wealth taxes, but in my own mind, it seems significant that a number of European countries that tried wealth taxes then repealed them. One common problem is that exemptions and exceptions start getting built into the law: for wealth held by the family farm, for wealth in other family businesses, for historical properties, for family heirlooms, for works of art, and so on and so on. Another problem is how to value wealth that isn’t being bought and sold in the market, and yet another problem is what happens when wealth is held in other countries. I’ve discussed here some more practical alternatives to a wealth tax, like focusing on the rule that allows the value of assets to be “stepped up” at death, so that heirs do not pay tax on earlier capital gains (see here and here). Indeed, just rolling back the US estate tax rules to 2004–which was hardly a time of confiscatory taxation–could raise about $100 billion per year in revenue.

Discussions with the US about the American health care system often counterpose the well-known issues of the US health care system (high costs, lack of health insurance, unequal access to care, lower health outcomes than many other high-income countries) with a nebulous alternative often called “national health insurance, like all the other countries in the world” or sometimes “Medicare for all.” I refer to the alternatives as “nebulous” because the suggestion seems to be that these alternative plans offer all the health care demanded, with all or the vast majority of costs covered by someone else.

But more than half the recipients of Medicare, for example, now have their benefits managed by a “Medicare Advantage” or Part C plan, in which the government makes a risk-adjusted payment to a private health insurance company to provide care. The traditional Medicare plan, in which the federal government makes direct payments to providers, also includes insurance premiums, co-pays and deductibles, to the extent that many users of traditional Medicare also purchase a “Medigap” health insurance plan to cover the risk of high out-of-pocket payments and uncovered services. Most of those who name-check “Medicare for all” do not seem to be envisioning health insurance partially paid for by the government, together with out-of-pocket payments, and a continued central role for private health insurance companies. Moreover, the long-term projections under existing law for future Medicare payments over the next 75 years are a cool $60 trillion or so bigger than the long-term projections for Medicare revenues. It might be prudent to address the long-term solvency of Medicare as it current exists before touting it as a solution.

When it comes to broad statements about national health insurance, it’s also important to specify what national models you have in mind–because national health insurance systems vary substantially across high=income countries. The Commonwealth Fund has recently published a webpage offering “International Health Care System Profiles” for 31 countries. Many of the countries are high-income: the US, Canada, ten countries in Europe, Japan, Australia, Taiwan, Israel. But there also a selection of other large economies around the world: China, India, Brazil, Nigeria, Ethiopia, Indonesia, Phillippines, and others. Each country health care profile runs about 20 pages: that, is the Commonwealth Fund is trying to ride the thin line between a reasonably in-depth overview and information overload.

When comparing the health care systems of other countries to the US health care system, the most common comparisons I hear are other English-speaking countries, and in particular Canada and England. Both countries have 100% population coverage for national health insurance, and both spend about 11% of GDP on health care, as opposed to 18% of GDP going to health care spending in the US. On one hand, the lower health care spending in those countries seems attractive, allowing for additional consumption of other goods and services; on the other hand, I suspect that concrete and specific American proposal to cut total health care spending by one-third–to the UK and Canada level–would go down in flaming defeat.

Moreover, the Canadian and English systems are quite different from one another. In Canada, 69% of people have private health insurance to complement the public program; in the UK, only about 12% have private health insurance. In Canada, the national health care system is organized on the provincial level–rather like giving all 50 US states the lead authority for organizing health insurance–while the British system is nationally organized. As a result, determination of wages for health care providers looks different: most British physicians are on salary, with most specialists paid directly by the National Health Service, while a number of Canadian physicians still work on a fee-for-service basis.

In short, it’s not enough just to bewail the US health care and health insurance system. Other national models differ substantially among themselves. Indeed, Germany, Netherlands, and Switzerland have organized their national health insurance systems around individual-level health insurance, combined with regulated competition among insurers. A serious discuss of US health care reform has a number of lessons to learn from other countries, but the healtgh care systems of other countries differ substantially, and a serious discussion needs to dig into the details. The national-level profiles from the Commonwealth Fund are a useful starting point on the journey to literacy in the choices of health care policy.

I have made no secret of my deep skepticism about President Trump’s tariff policies since he announced his “Liberation Day” tariffs in April 2025. My first-blink “Thoughts on the Trump Tariffs” from a few days after the announcement has held held up reasonably well, and I have posted follow-ups on the subject a number of times since. For an overview of economic thinking on the subject, I recommend the six-paper symposium on tariffs in the Summer 2026 issue of the Journal of Economic Perspectives (where I work as Managing Editor).

Of course, I’m sure that all readers of this blog have already scrutinized these papers at great depth. But on the off chance you were pursuing other plans when the issue was published in August, it seemed worth mentioning here. I will not try here to summarize the papers in depth, but instead just to tell you something interesting about each one.

In “US Tariff Policy since 1789,” Miguel Acosta, Lydia Cox, Andrew Greenland, John Lopresti, Christopher M. Meissner, Martin Rotemberg, and Sharon Traiberman describe their efforts to construct a detailed database of tariffs back through 1789–a monumental task which involves classifying products in a consistent way over long periods of time and then tracking down tariff rates as they were published in 19th century Congressional documents. They emphasize the three R’s of historical tariff policy: “the use of tariffs as a key revenue stream prior to the Civil War; the restrictive tariffs of the newly emergent industrial economy after the Civil War; and the period of “reciprocity” that emerged beginning in the 1930s.” I was intrigued by how tariff policy has over time come to focus on very specific goods (say, types of cheese) as a way of focusing tariffs on some countries rather than others. Here’s a graph showing the number of products with tariffs on them over time–a number that keeps rising as the tariffs become more detailed.

In “Labor Market Responses to Tariffs: Frictions, Dynamics, and Policy Responses,” Rafael Dix-Carneiro and Brian K. Kovak present evidence on patterns of jobs lost to trade over time. They argue: “A central finding is that trade liberalization and import competition can generate sizable and highly localized labor market disruptions. Specifically, employment and earnings losses tend to be concentrated in particular regions, sectors, and occupations, and adjustment is often slow, with effects extending beyond directly affected workers to entire local communities.” In analyzing patterns of adjustment, they show that the costs fall especially heavily on older workers with less flexibility to shift their lives. They also find that when industries are protected from import competition by tariffs, they often react by investing more in capital equipment and by earning higher profits, rather than by keeping employment high. They make a case for “smoothing policies,” to cushion the costs of trade shocks for those bearing the highest costs, with the goal of helping to facilitate the longer-run benefits of trade for the economy as a whole.

In “Should We Tax Trade? A Pigouvian Perspective,” by Arnaud Costinot and Iván Werning point out that it is straightforward to draw up a model showing that that those who create the disruptions of trade should also pay for the social costs–but that real-world trade policy is also not so simple. They suggest the “Targeting Principle”: “The economy may be distorted and government intervention may be warranted. Yet one cannot conclude that a tariff is optimal without making it compete with other available instruments. The Targeting Principle suggests that trade policy is acupuncture with a fork and calls for using needles instead. … Let us return once more to the case where domestic redistribution is the only motive for trade policy intervention. On the one hand, it is unrealistic to assume that no other policy instruments are available to achieve this goal. As Rodrik (1995) puts it, “Saying that trade policy exists because it serves to transfer income to favored groups is a bit like saying Sir Edmund Hillary climbed Mt. Everest because he wanted to get some mountain air. There was surely an easier way of accomplishing that objective.” The takeaway is that arguments for specific tariff policy focused on narrow geopolitical or economic goals are likely to be more persuasive than sweeping and general pro-tariff arguments.

In The Incidence of Tariffs: Rates and Reality,” Gita Gopinath and Brent Neiman dig into the question of whether tariffs are paid by foreign producers–as President Trump promised in April 2025–or whether they raise prices for US consumers. They point out that through the end of 2025, the actual tariff rates paid were only about half of the announced statutory rates, because “[s]hipment lags, product and company-specific exemptions, utilization of the US-Mexico-Canada Agreement (USMCA), and evasion and heterogeneous enforcement all contribute to the large gap between statutory and actual rates.” With these kinds of factors taken into account, their estimates are that more than 90% of the tariffs were passed though to US buyers–a finding that they point out is consistent with a number of other studies.

In “Global Imbalances, Tariffs, and Industrial Policy, Pierre-Olivier Gourinchas, Gene Kindberg-Hanlon, Manasa Patnam, Lorenzo Rotunno, and Michele Ruta work through the question of tariffs or industrial policy in general can reduce trade deficits. They re-affirm the conventional (among economists) argument that tariffs alone are unlikely to have much affect on trade deficits, because trade deficits are a macroeconomic outcome driven by domestic savings and invesment levels. Similar logic applies to targetted industrial policy. However, what they call “macroeconomic industrial policies” involving financial repression and forced savings raise different issues. They write:

The term “financial repression” refers to a group of policies. It may involve limiting the interest rates available on deposits in the domestic banking sector or limiting the availability of other financial assets, in order to reduce funding costs, while simultaneously limiting lending rates on loans to businesses in targeted sectors. These policies are often combined with forced saving policies that can include prohibitions on dividend payments by corporations (to encourage higher retained earnings), reserve requirements on export earnings, mandatory saving requirements in pensions or similar products, or even weak social safety net provision that requires high private saving for insurance purposes.

The goal of rules and incentives that drive up saving and limit investment options for households is to make it cheaper and easier for domestic firms to raise investment capital. China is of course a major example of such policies. By driving up domestic savings and investment, these policies can lead to trade surpluses–but a policy of lower consumption and limited financial investment opportunities is unlikely to be a political winner for any US politician.

In Evaluating the Fiscal and Distributional Implications of Tariffs,” Kyle Pomerleau and Erica York look at how prominent organizations like the Congressional Budget Office, the Tax Policy Center run by the Brookings Institution and the Urban Institute, the Yale Budget Lab, the Penn/Wharton Budget Model, and the Tax Foundation formulate their stimates for total revenue to be raised by the tariffs (at least as they existed in 2025), and also how some of these organzations formulate estimates of who pays for tariffs across the income distribution. On the first question, tariffs at more-or-less the 2025 levels would raise about $2.5 trillion in revenue over the next 10 years–that is, an average of $250 billion per year–or someone less if the model takes into account how tariffs will slow the economy. In distributional terms, the tariffs tend to weigh more heavily on middle-income and low-income households, and less on those near the top of the income distribution.

As I wrote back in April 2025, President Trump owns the tariffs. He imposed them without Congress passing any new laws, and when the courts have told him that he did not have the legal power to do so, he has then offered one different legal rationale after another. He made extravagant promises about how the tariffs: for example, when President Trump was talking about the tariffs to come, he said: “All I know is this: We’re going to take in hundreds of millions of dollars in tariffs, and we’re going to become so rich, you’re not going to know where to spend all that money! I’m telling you – you just watch. We’re going to have jobs, we’re going to have open factories, it’s going to be great.” As I wrote in April 2025: “President Trump’s claims about the benefits of tariffs seem based on demonstrably false beliefs. … I confess to a darker thought. Part of me hopes that Trump will keep his tariffs in place until the costs are broadly apparent to all, so that a modern consciousness of why this approach doesn’t work can take effect for the next few decades.”

For many low-income countries, perhaps the greatest deficit they face is a lack of the human skills that build economic growth. The new artificial intelligence tools improve access to a wide range skills. Thus, the World Bank devoted its flagship World Development Report 2026 to “The Promise of Artificial Intelligence.”

For a vivid example, consider agricultural extension agents, who are goverment-funded to travel across rural areas providing insight into how output might be improved. The WDR writes:

The planning guidelines of the Food and Agriculture Organization of the United
Nations (FAO) for agricultural extension services specify one agent for every 500–2,000 rural inhabitants, depending on a country’s production systems and the density of its population. But the reality on the ground falls short. Recent estimates indicate that in Nigeria, for example, a single public extension agent may serve anywhere from 5,000 to 10,000 farmers. In India, which has one of the largest public extension systems in the world, with about 120,000 extension professionals—the ratio of extension workers to farmers is about 1 to 5,000 … A gap in service delivery is evident not only in the number of extension agents available to farmers, but also in the quality and responsiveness of the advice they provide. Public extension advice tends to be structurally generic (designed for an average
farmer in an average season) …

If local farmers have some access to a smartphone in the nearest village, and can query an AI tool about local conditions, crops, and what’s working for others, including the ability to send along pictures of the landscape and how the crops are faring, it can make make the difference between extreme poverty and malnutrition in a given year–or not. One experiment with AI advisors in Ethiopia providing locally customized advice increased wheat yields 38%.

As the report points out, low-income countries are unlikely to be significant players in the near term in developing frontier cutting-edge AI models, or in producing the most sophisticated semiconductor chips. But in a number of contexts, the use of AI often benefits those starting with less experience or skill, because those workers, firms, and government have the most room for gains from greater access to knowledge and skill. A number of companies and governments in lower-income countries are already using these tools, including use of chatbots for questions and feedback on public services and from the customers of firms. Small AIs can be “built for one narrow task, such as identifying a crop disease or screening a medical image.” The report notes:

Many developing countries face shortages of highly skilled workers. AI can help by enabling less experienced workers to perform more advanced cognitive tasks, allowing some work to be shifted away from scarce specialists and making existing workers more productive. First, if AI automates more expert tasks within an occupation, it lowers the entry requirements
for people to carry out the work in that occupation.There is evidence of such leveling up across cognitive occupations, including customer support,professional writing, and software development. Second, AI can shift a given cognitive task to adjacent occupations that employ people with less expertise, such as shifting the task of medical-image screening from scarce medical specialists to lower-level health care providers. Third, AI makes people more productive at what they currently do. For example, generative AI tools improve the quality of instruction in schools by supporting teachers with lesson planning, generating learning materials matched to students’ skill levels, and providing quicker feedback on student work.

The WDR is open about acknowleding a range of practical issues. AI will also disrupt some industries in lower-income countries: for example, AI agents seem likely to displace a number of workers at call centers. Benefits from AI require complementary investments in physical and human infrastructure. Low-income countries may wish to have a process for adaping existing open-source AI tools for local use and/or foor designating certain AI tools as trustworthy. The use of AI in schools can be a decidedly mixed blessing, but the possibilities for customized tutoring when teachers are scarce seem substantial.

Living in the environment of US higher education, as I do, one hears a lot of concern about the use of AI in writing research papers, or in writing novels. A chief concern is that the use of AI involves a loss of something precious and ineffably human. Those issues matter. But the WDR reminds us that the potential gains and tradeoffs may look quite different in low-income developing economies.

The amounts of money being spent on the hardware and buildings behind the new AI boom may turn out to be enormously large by historical standards. How is it being financed?

Stijn Van Nieuwerburgh tells the story in “Financing the AI Buildout” (Brookings Papers on Economic Activity, Fall 2026, conference draft version). As he points out, the new AI technologies are often discussed as a new form of software, and how the capabilities of that software are developing. The software perspective is true, but partial. The new AI technologies are also a physical capital boom. Indeed, our public discourse on AI seems to have entered a period where one set of concerns is that Ai software will be so cheap and available that it will lead to mass unemployment, while another set of concerns is that building the physical capital behind the AI boom–including data centers, semiconductor plants, additional energy production, and more–will be extraordinarily expensive. I’m alwas loathe to make predictions, but I feel confident in saying that the new AI technologies will not be both cheap and expensive at the same time.

Van Nieuwenbergh walks through a set of calculations to forecast the size physical capital AI investment in the next few years, based on current plans and the recognition that not all the current plan will actually happen. He produces this historical comparison, which needs to be taken with a few grains of salt. The AI buildout is a prediction, while the others actually happened. Also, the time periods vary: the railroad investment, the second-highest number on the right-hand column, lasted much longer than these predictions for the AI buildout. Still, the estimates for the cost of the AI buildout are striking.

Where are firms getting the funding for this buildout? Remember, stock market wealth is not an immediate source of funding–unless you sell the stock to someone else. One of the economic puzzles of the last decade or so is that big US tech companies were holding huge amounts of cash, more than $1 trillion (for example, I wrote about this back in 2013, here and here). A common question has been: What investment opportunity were they waiting for? The answer appears to be the AI buildout. Van Nieuwenbergh writes:

Aggregate capital expenditures by Oracle, Microsoft, Amazon, Meta, and Alphabet rose from roughly $97 billion in 2020 to more than $400 billion in 2025 and are projected to exceed $800 billion in 2026. On current estimates, capex will surpass the five firms’ combined operating cash flow for the first time in 2026. The figure therefore marks an important transition: internal cash generation remains substantial, but it is no longer sufficient to finance the projected pace of investment without greater reliance on external capital or financing structures that shift assets and obligations away from the operating companies’ balance sheets.

The new outside financing is a combination of equity and debt finance. The equity finance is arriving in the form of “[s]pecialized data center developers, REITs [real estate investment trusts], infrastructure funds, and private equity investors.” The debt finance includes bank loans and bonds, but private credit funds provide a growing share.

For those with an interest in private credit in particular, Puriya Abbassi, Iñaki Aldasoro, and Sebastian Doerr provide an overview of “Financing the digital economy: the role of private credit “ (BIS Quarterly Review, September 2026, pp. 17-30). In contrast to bank loans, which often require that a borrower have physical collateral to back up the loan, private credit funds make loans based on expected cash flows and intangible assets. As banks have become more tightly regulated, private credit funds have expanded, and as of 2025 had loan amounts outstanding of $2.5 trillion. As the authors note: “Aggregate [private credit] borrowing by tech firms rose from around $22 billion (22% of total private credit) in 2010 to over $1 trillion (44%) by 2025.” These private credit funds typically operate by getting money from investors that is usually committed for the middle-term of five to eight years, and cannot be withdrawn. If a bank has many loans that go sour, the bank has been lending out money of depositors, which can cause a problem; if a private credit fund loan portfolio goes sour, the investor in that fund would suffer losses, but there is much less reason to think that those losses would propogate into banking or the rest of the financial system.

One of the strengths of a market economy is that some companies can succeed, while others fail, without endangering the health of the economy as a whole. But when the scale of the investment gets this big, and the risks are distributed across a web of interlocking companies, it’s natural to wonder about which investors will bear the losses if the AI boom doesn’t generate revenue as fast as expected–and whether losses from those investors might be substantial enough to affect the broader economy. Some readers will remember the dot-com boom-and-bust when the overall stock market fell by about 50% from March 2000 to October 2002, and many internet companies fell by even more. There was a recession for eight months in 2001. But although the economy took a hit and a number of companies vanished outright, the technology of the internet didn’t go away, and the physical capital boom behind the internet was put to productive use.

In some ways, the physical capital boom surrounding AI puts me in mind of a rhetorical question I remember hearing about the California gold rush in the mid-1800s: During the California gold rush, which group got richer: the gold miners, or the people who sold picks and shovels to the gold miners? Many of the gold miners were unsuccessful, so the average gain for a gold miner wasn’t all that high. But every gold miner, successful or not, needed to buy the tools. Right now, the stock market prices of AI software companies are sky-high. But some of those companies are likely to become unsuccessful goldminers–who nonetheless had to pay the costs for the physical investment in the new technologies.

Economists and policy-makers know, in a broad sense, how to respond to a negative shock to aggregate demand in an economy, like the financial meltdown leading to the Great Recession of 2007-09. The financial crisis shrinks demand in the economy, recession hits, and unemployment rises. The government can play a role in softening and shortening the recession through expansionary fiscal policy (spending increases and tax cuts) or monetary policy (lower interest rates). To be clear, this policy recipe is not about building robust long-term economic health, which would require thinking about education and job training, incentives for capital investment, research and development of new technology, and balancing the benefits and costs of regulation. But softening and shortening the immediate blow of recessions is a worthwhile objective.

But when the shocks that hit the economy are about about aggregate supply, like the pandemic recession, the policy prescriptions based on responding to demand shocks don’t apply as well. The key difference is that with a negative shock to aggregate demand, inflation is naturally tamped down, so when the government acts to stimulate the economy, there is little risk of inflation taking off. But in a negative shock to aggregate supply, unemployment rises and prices also start rising. Thus, if the government stimulates the economy, it can reduce immediate hardship, but at risk of driving inflation higher. Thus, there’s no inflationary pressure after the Great Recession, but there certainly was after the pandemic recession.

Peter R. Orszag, Eric Van Nostrand, Noah Barbieri, and Layla McDermott make a case that the US and world economy are experiencing a growing number and risk of supply-side shocks in “The New Supply-Side Era: The O-Ring Meets Geopolitics” (September 2026, also, a video of Orszag presenting the paper at the London School of Economics is available). They argue that the global economy has now experienced six supply-side shocks in the last six years: “[t]he pandemic, artificial intelligence, Russia’s invasion of Ukraine, tariff shocks, China’s rare-earth export controls, and the war in the Middle East.”

In addition, they add a complicating factor, what Michael Kremer (Nobel ’19) called the O-ring theory of development. After the space shuttle Challenger broke apart after liftoff in 1986, the study of what caused the crash focused on the O-rings, a rubber gasket used to seal a connection between two pipes. Apparently the rubber gasket became brittle when unusually cold, but this feature was not well-known at the time. The key insight of the O-ring theory of development is that the performance of a complex piece of machinery can be limited by the weakest component–even when everything else is going just fine. The pandemic revealed various O-rings in the US economy, like the cargo ships sitting offshore and waiting to offload. The authors write:

One weak link can reduce or destroy the value of everything else. That multiplicative structure is what makes today’s economy so exposed to supply shocks: as products became more complex, a generation of globalization built long supply chains, and a problematic input at one stage—a chip, a mineral, a precursor chemical—can shut down production far beyond it. These supply shocks behave quite differently from the textbook demand cycle. Their effects develop slowly, propagate through supply chains in waves—delivery delays, then input shortages, then margin compression, then price pass-through—and unwind with long and uneven lags. … What distinguishes the new era is thus not necessarily the blast radius of any single shock, but instead their frequency and overlap. Disruptions (especially in energy) may tend to have smaller effects than those of their historical counterparts, but they arrive more often and, increasingly, on top of one another.

To put it another way, the prevailing gestalt a few decades back was that truly important supply shocks were relatively rare. The modern a parade of supply shocks coming thick and fast in a more interconnected global economy–where some of the interconnections impose O-ring constraints and costs.

Orszag, Nostrand, Barbieri, and McDermott also emphasize that many of these O-ring constraints are physical or geographic: shortages of key components, the capacity of ports, an underwater seabed cable severed, a chokepoint for the passage of shipping, a country with most of the world’s capacity to refine rare earth materials (China) or producing the highest-end computer chips (Taiwan), or even tanks and planes crossing a border. A string of six major supply-side disruptions in six years, at times interrelated and reinforcing each other, starts to look less like “bad luck” and more like “the way things are going to be.” The authors argue that economists and economics are being pushed into a world where we need to think as “geopolitical supply-siders,” attuned to the prevalence of supply-side shocks and how they propagate through an economy.

The pandemic recession was a classic

The Medicaid program spends close to $1 trillion annually, with about 65% of that coming fom the federal government and the rest from state governnments. This is almost, but not quite, one-in-five dollars of total US health care spending. Medicaid is not quite as big as Medicare, which is about $1.2 trillion per year, but Medicaid alone is almost one-in-five dollars of total US health care spending. James Hamlette and Karen Kopecky offer a check check-in on the Medicaid program in “Who’s Driving Medicaid’s Enrollment Growth?” (Economic Commentary: Federal Reserve Bank of Cleveland, October 13, 2026). As they write about Medicaid:

[I]t finances primary and acute medical care and long-term care for eligible low-income individuals. The criteria for eligibility have changed over time, and today, eligible populations span a wide range, including children, pregnant women, low-income elderly and disabled individuals, and working adults in low-wage jobs. The program is jointly administered and funded at federal and state levels. It requires states to provide specific benefits—including inpatient and outpatient hospital services, physician services, nursing home services, mental health services, and preventative care for children—with the option to cover additional services such as prescription drugs, home- and community-based services, and dental care. Healthcare services have grown over time as a proportion of US consumption spending and gross domestic product, and Medicaid outlays per capita have reflected this trend (Congressional Research Service, 2026), increasing steadily in recent decades and reaching four times their 1990 level in 2024.

As the authors emphasize, the main driver behind Medicaid spending growth is that many more people are enrolled in the program. In 1990, about 10% of all Americans were in Medicaid: now, it’s approaching 30%. Indeed, average spending per person in Medicaid–adjusted for the overall rate of inflation–has barely budged since 1990.

Hamlette and Kopecky dig into the Medicaid enrollment details and show: “The main drivers of this coverage expansion have changed over time. Prior to 2000, enrollment growth was driven by an increase in coverage of children and their families. From 2000 to 2013, enrollment growth was driven by an increase in coverage of childless nondisabled and disabled adults,² with similar increases in the number of working and nonworking adults on Medicaid. Since 2014, the Affordable Care Act (ACA) Medicaid expansion has driven growth, with expansion states extending coverage to adults earning up to 138 percent of the federal poverty level, and enrollment growth during this period has been driven primarily by working adults.”

But in thinking about Medicaid, it’s important to remember that different groups can have very different health care expenses. For example, the disabled and the elderly together are 17.3% of the Medicaid population. but just over half of all Medicaid spending.

The main current proposals for holding down Medicaid costs seem to involve requiring that Medicaid recipients work. But given that Medicaid spending on non-disabled, non-elderly adults is only about 15.8% of the total–and many of them are already working–the cost savings from this approach are likely to be limited. A more productive approach might be to dig into how different states have designed their Medicaid program. Mark Shepard and Jacob Wallace dig into these issues in “Understanding Medicaid Managed Care: The Procured Competition Model” in the Spring 2026 issue of the Journal of Economic Perspectives (where I work as Managing Editor). As they note in their abstract, they “discuss the key policy levers across procurement, competition, and consumer choice in this model. Throughout, we highlight open research questions, arguing that the enormous variation in how states design these programs—combined with limited evidence on what works—represents a promising area for high-impact scholarship.”

Global population is expected to keep rising for the next 50-60 years, because it will take awhile for the demographic force of declining birthrates to overcome the life expectancies. In addition, the world still has lots of very poor people: for example, about 800 million people around the world live below the World Bank extreme poverty line of $3/day, and of course hundreds of millions more are not far above that line. With a rising world population, and one where we hope the standard of living for the poorest will be rising, more food production will be needed. So how is the world doing in this area?

The Global Agricultural Productivity initiative at Virgina Tech University,tackles the big questions of global agricultural productivity in the 2026 GAP Report, Restarting Growth: The Role of Regenerative Agriculture in Productivity Growth and Profitability, with a team effort led by Chellie Maples. Here’s a picture to set the stage:

Growth in total agricultural output can come from several sources: more land, more irrigation, more intensive use of inputs (labor, materials, and capital), and increased productivity given the same amounts of land, irrigation, and inputs (say, from improved farming practices, better weather information, improved seeds, and so on). To do these calculations, gains in agricultural output that cannot be attributed to measurable factors like land, irrigation, and inputs are then ascribed to productivity gains.

In the 1960s, most growth of agriculture came from more intensive use of inputs. By the 1990s and early 2000s, most of the growth of agricultural output was coming from producitvity gains, with very little coming from land expansion or additional inputs. But since about 2011, growth of agricultural output has slumped. Productivity gains are way down, although expansion of land and inputs have made up some of the difference. In a nutshell, this is global challenge of finding ecologically-sound ways of increasing agricultural output.

Of course, these factors vary by region. But even in the areas with relatively rapid growth in agricultural output, that growth is being driven by additional inputs (like South Asia) or by additional land (sub-Saharan Africa).

Part of the answer here is to improve productivity growth in agriculture, especially in the regions where economic and population growth are creating stresses on the food supply. I’ve argued in the past for a rise in global R&D spending in agriculture, especially focused on lower-income areas of the world. (Also, here’s a look at the long-term prospects for agricultural productivity growth from about 25 years back by Vernon Ruttan, a preeminent agricultural economist.)

The particular focus of this year’s GAP report is on “regenerative agriculture,” defined as “a set of contextually appropriate farming and management practices and tools oriented toward improving soil and ecosystem health as a primary goal, with enhanced biodiversity, watershed function, and resilience as downstream benefits. Which practices qualify depends on local soil, climate, and production system conditions. This definition aligns practices, principles, and outcomes, centering ecosystem health as the mechanism through which broader gains are achieved.” In my own mind, this focus starts from the hard fact that additional inputs and land are driving growth of total agricultural output, recognizes that expanding agriculture to marginal and heavily fertilized land will have environmental tradeoffs, and then uses this lens to think about productivity gains that could minimize these environmental consequences. The possibilities in this area seem real, but the report is forthright in noting:

Regenerative agriculture (RA) has drawn growing attention, in part because its practices target the input use and natural capital depletion that the current output growth model largely depends on. Yet the potential contribution of RA practices to TFP [total factor productivity] growth has, thus far, only been a conceptual claim rather than an empirical finding. No study in the current literature directly measures changes in TFP growth attributable to RA.

Sometimes you see a figure that makes you uncertain whether to laugh or cry, so you do a little of both. The Global Linkages Lab at Brown University–specifically, Şebnem Kalemli-Özcan, Can Soylu, and Muhammed A. Yildirim–have sought to track down “every occasion on which the President attached a number to a US tariff, whatever the channel: speeches, interviews and press gaggles, Truth Social posts, executive orders and proclamations, and White House fact sheets and statements.” Some announcements were made more than once in different formats, so their list from January 21, 2025 up to the present includes 749 total such statements, covering 341 separate tariff changes. They call it TiRADE, for “Tariff Rates in Announcements, Declarations, and Events.” Blue dots show threatened or announced tariffs, green dots show implemented tariffs.

Details behind the figure and sources for the statements are all available at the TiRADE website. Here, I just want to emphasize the uncertainty and instability of it all. I would probably disagree with an overall plan to raise tariffs in broad ways, and I would have to think through the implication of raising tariffs in narrow and targeted ways on a case-by-case basis. But this figure makes it transparent that the Trump-dictated US tariff policy since early 2025 has not been a coherent plan, but rather a sputtering and erratic seat-of-the-pants shambles.

The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2026 has been awarded to Robert E. Hall and Michael Woodford “”for microeconomic foundations of macroeconomic stabilisation.” To gain some intuition for what that actually means, it’s useful to flash back a few decades and remember the origins of “macroeconomics” and how macroeconomic analysis was done a half-century ago. There’s an old phrase about “painting with a broad brush,” when you skip past smaller details. In this brief overview, I intend to go well beyond painting with a broad brush, perhaps closer to painting with an industrial-power spray-gun.

Economists have been writing about subjects related to what we now call “macroeconomics” for centuries. But the actual terminology seems to have emerged during the Great Depression, when it became abundantly clear that entire national economies could malfunction at the same time without the occurence of a disaster like a war or pandemic. The great Norwegian economist Ragnar Frisch, co-winner of the first Nobel prize in economics awarded in 1969, was inventing a new terminology when he wrote in 1933: “”The micro-dynamic analysis is an analysis by which we try to explain in some detail the behaviour of a certain section of the huge economic mechanism, taking for granted that certain general parameters are given … The macrodynamic analysis, on the other hand, tries to give an account of the whole economic system taken in its entirety.” By the 1940s, the terms “microeconomics” and “macroeconomics” were being commonly used in the professional writing of economists, and even in a few textbooks.

At about the same time, not coincidentally, Simon Kuznets was leading an effort to measure the entire economy–that is, he was defining and inventing what we now refer to as “gross domestic product.” Kuznets received the Nobel prize in 1971, the third year it was awarded. As it became conceivable to write down what was meant by a macroeoconomy, in categories, Wassily Leontief created the “input-output method,” in which you conceive of the economy as a giant spreadsheet, with all economic inputs listed as columns and all economic outputs listed as rows–thus showing what inputs went into each output. Leontief wins the Nobel prize in 1973.

Looking at an economy as an organized group of inputs and outputs combined easily with practical questions: By looking at these relationships in the past, can we forecast what will happen in the future if there is, say, a boom or bust in one area of the economy? Can we use this tool for public policy, perhaps to prevent a recurrence of the Great Depression? In 1980, Lawrence Klein won the Nobel prize for work in building up many-equation macroeconomic models: for example, his famous Wharton model in the late 1960s started with 200 equations, but eventually expanded to more than 1,000 equations. The macroeconomic model had joined with the then-newfangled power of computers.

There was a brief shining moment for economists in the 1960s when it seemed as if all of this was leading to a full understanding of the macroeconomy and appropriate macroeconomic policies. And then came the deep recessions, unemployment, and high inflation of the 1970s, and all that confidence went out the window. Among the top concerns was that there had been an an observed “Phillips curve” relationship in in the preceding decades that if unemployment was extremely low, it was a sign that the economy was “overheating” and that inflation was likely to rise, but if unemployment was high, inflationary pressures would then drop off. But the 1970s famously had “stagflation”–both high unemployment and high inflation at the same time. Such an outcome was literally impossible in the context of the big macroeconomic models of the 1960s.

The field of economics is never short of skeptics, and there were skeptics about the 1,000-equation macroeconomic models, too. However, many of those skeptics tended to have proposals for keeping the models and redesigning them in some way. The 1978 paper by Robert Hall cited by this year’s Nobel prize committee offered a much more fundamental challenge.

Consider the problem of predicting changes in consumption, which makes up about two-thirds of aggregate demand in the macroeconomy. The conventional models of the 1960s viewed consumption as determined by current (or recent) income: when people got income, they had a “marginal propensity to consume” some of it and to save the rest, in a way that followed past patterns. In contrast, Hall argued that people made consumption decisions looking forward into the future: that is, choices about how much to consume and how much to save were based on beliefs about whether jobs seemed likely to be secure in the future, whether there were risks of recession or inflation, and other factors.

There was a clear empirical test to differentiate these approaches. The conventional approach made predictions about changes in consumption based on the the 1,000-equation model. But Hall reasoned that if current consumption choices were based on future beliefs, then changes in consumption would be based on changes in those future beliefs. By definition, no one can say how beliefs are going to change in the future (if you could, it wouild already be part of your current beliefs). As a result, Hall argued that changes in consumption should be unpredictable and random; in mathematical terms, a random walk. And by golly, when you compared predictions of consumption based on the 1,000-equation models to a prediction of consumption that it was a random change from the prevous period, it turned out that the giant multi-equation model was no better than the random walk.

Hall’s challenge (along with related work done by others) was a hammer-blow to the old-style macroeconomic theorizing. At a deeper level, the challenge pointed out that the 1,000-equation macroeconomic models were snapshots of what was happening in the economy at a certain time, which could be useful. But these snapshots did not explicitly include consumers making decisions, or firms making decisions, or a central bank making decisions. An alternative framework was already under development in the 1970s, which in turn led to Nobel prizes for Robert E. Lucas in 1995, for Finn E. Kydland and Edward C. Prescott in 2004, and for Edmund S. Phelps in 2006, as well as many other contributors who didn’t win a Nobel.

But in some ways, this new version of macroeconomics had lots of individual insights, but not a compact overall perspective, on which more complex formulations could be built. This is where Michael Woodford enters our scenario. Each year, the Nobel prize committee writes a “Scientific background” paper to explain the contributions of the prize-winners, and I quote from it here:

A new framework for macroeconomic stabilization policy needed to be built up, one that, like the Keynesian approach, also recognized the shortcomings of markets and, therefore, the need for government interventions. However, in contrast to the previous framework, the new one had to rest on an understanding of the fact that macroeconomic outcomes cannot simply be represented as stable relationships between aggregate variables. Rather, macroeconomic outcomes had to be seen as the cumulative result of actions taken by all the individual consumers and firms in the economy. The experience during the 1970s also convinced the profession that consumers and firms were far more forward-looking than had been previously thought: they pay careful attention to where both monetary and fiscal policy are heading and, hence, to expected inflation and future aggregate activity. Therefore, a new framework had to describe economic actors as purposefully reacting to government stabilization policy as well as to the fundamental factors causing the fluctuations in economic activity. Lastly, these fundamental factors should be expected to vary in nature over time: different recessions occur for different fundamental reasons, and may therefore also require different policy measures. …

Through a series of subsequent papers, built on Hall’s consumption framework as well as other important developments in macroeconomic theory during the 1970s and 1980s, Michael Woodford arose as the key architect behind the so-called New Keynesian model: a full macroeconomic stabilization framework that has become the main tool for policymakers today. In his 2003 monograph, Woodford summarizes as well as further develops the general setting. The original theory was chiefly aimed as a foundation for monetary policy analysis but subsequently also became the main tool for the analysis of fiscal policy. This model features departures from perfect competition, with firms charging markups over marginal cost, as well as sticky prices and wages preventing the efficient use of resources. Despite allowing consumers and firms in the model to adapt their behavior to changes in government policy, it shows how appropriately used fiscal and monetary policy can improve welfare by stabilizing the economy in the face of disturbances. …

The simplest New Keynesian model has only three equations; nevertheless, it is a very powerful tool for understanding some key aspects of how monetary policy works. The first equation summarizes consumer behavior and is in fact Hall’s theory of consumption. The second equation describes how profit-maximizing firms set prices when it is costly to change them, and the third and final equation captures the central bank’s interest-rate policy. When analyzing monetary policy quantitatively using this framework, additional variables and equations are added; central banks often have medium- as well as large-scale versions tailored to their national economies in question. Similarly, finance ministries and government agencies analyzing fiscal policy build versions of the model with yet another, slightly different, focus.

Taking this together, macroeconomics has undergone an enormous pivot in the last half-century. The macroeconomics of the 1960s was (implicitly) built on a belief that if one could write down a detailed version of what was happening in the macroeconomy for a given period, like a year, one could then rely on the relationships between the main aggregate economic variables to be roughly constant over time, in a way that would allow analysis and policy-making. But this vision of macroeconomics lacked expectation-forming consumers, firms, and policy-makers, interacting in an economy with imperfect competition and slow-to-adjust (“sticky”) prices and wages. In a phrase, it lacked “microeconomic foundations.” The task of building out alternative versions of Woodford’s very basic approach is still very much a work in progress: as one of many examples, you will notice that the basic three-equation model doesn’t have a financial sector. For a starting point on how economists, policymakers, and central banks have been building on this model over time, I recommend both the “Scientific Background” paper from the Nobel prize committee and Jordi Galí’s article “The State of New Keynesian Economics: A Partial Assessment,” from the Summer 2018 issue of the Journal of Economic Perspectives (where I work as Managing Editor.)