Conversable Economist

The world economy is now about a year into President Trump’s assault on the global trading system. Richard Baldwin reviews what has happened and suggests where it is all headed in World War Trade (Centre for Economic Policy Research, 2026), a long essay in the form of a short online book. As Baldwin notes: “The working assumption is no longer that trade and investment are safe by default, underwritten by American leadership and Chinese growth. The new assumption is that while World War Trade may go quiet for long stretches, the weapons will remain deployed. … The old assumption that globalisation is `safe by default’ is gone, permanently.”

Baldwin offers some reminders of how US leadership used to talk about international trade, referring back to a speech from President John F. Kennedy when he signed the Trade Expansion Act of 1962. Kennedy said:

“Today I am signing the Trade Expansion Act of 1962 … It marks a decisive point for the future of our economy, for our relations with our friends and allies, and for the prospects of free institutions and free societies everywhere. … This act recognizes, fully and completely, that we cannot protect our economy by stagnating behind tariff walls. … The best protection possible is a mutual lowering of tariff barriers among friendly nations so that all may benefit from a free flow of goods. … Increased economic activity resulting from increased trade will provide more job opportunities for our workers. … Our industries will be stimulated by increased export opportunities … The results can bring a dynamic new era of growth.”

Trump’s tariffs have run into the problems predicted when they were imposed. One promise was that tariffs would supercharge US manufacturing. But in fact, US manufacturers have suffered under Trump’s tariffs. as Baldwin writes:

Donald Trump’s tariff theory reflected a misunderstanding of how American manufacturing worked in the 21st century. American firms do not simply make cars, planes or refrigerators. They manage elaborate international supply chains to assemble high-quality products at competitive prices. Today, you have to import to export.

Another promise was that the tariff would all be paid by foreigners. But US consumers began to notice that the higher prices were showing up on US retail shelves:

The Trump administration seemed to be genuinely surprised when the Affordability Crisis emerged. The President had long believed that foreigners would pay the tariffs. This belief, based on a classic economic fallacy, was core to the “foundation myth” on which the entire Trumpian trade policy was based. … [In November 2025], Donald Trump lowered tariffs on more than 200 products that US consumers routinely purchase to feed their families at home, many of which had seen double-digit price increases since Trump won in November 2024. The items read like a grocery list: coffee, tea, cocoa, bananas, oranges, tomatoes and beef.

We have been living through the silly season of Trump’s tariff policy for some months now. Baldwin lays out the details: here’s my own summary. The Trump administration has made innumerable announcements about tariff policy, and you will be stunned to learn that every single one of them is a greater triumph than the one before, natch. High announced tariffs? A triumph. Announcing an agreement that would reduce those tariffs? Another triumph. Creating exceptions and loopholes in the lower tariffs to ease the pain on US consumers and on US firms importing inputs to production? Yet another triumph. Announcing a new round of high tariffs? One more triumph. A new tariff policy has a completely different reason than the previous tariff policy? Yet another triumph of statesmanship. Indeed, every time a previous tariff policy is changed, or even abolished, it simply demonstrates that all previous tariff policies were triumphs. Then the US Supreme Court ruled that most of the tariffs imposed since April 2, 2025, were all unconstitutional to begin with. And President Trump reacted by imposing yet another round of tariffs with another pretextual legal rationale.

As US manufacturing firms struggle to deal with higher prices and cutoffs and heightened uncertainty of their global supply chains for inputs, and US consumers face higher prices as a result of tariffs, what’s the rest of the world doing? Baldwin argues persuasively that other nations of the world are pursuing regional free-trade agreements that pointedly leave out the United States–so that US firms have no voice in the negotiations. Baldwin calls it the “domino theory of regionalism,” which is the idea that regional free trade agreements benefit those who are inside, and thus disadvantage those who are outside. Every time an outsider decides to join up, it’s one more domino falling into place. As one example of these growing regional ties (and there are a number of such examples), Baldwin write about the 27 member states of the European Union, the 12 countries belonging to the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP), and the 11 countries that are members of the Association of Southeast Asian Nations (ASEAN):

The EU, CPTPP and ASEAN are already working toward closer cooperation, having launched “Trade and Investment Dialogues257” in November 2025. The Dialogues produced first steps towards cooperation including statements of commitment to a free, open, and rules-based trading system as well as an agreement to focus future talks on supply chain resilience, digital trade, investment facilitation, and upholding WTO rules. No binding agreements were concluded, but the Dialogues will be a platform for ongoing engagement.

Baldwin writes of Trump’s “Liberation Day” tariffs announced on April 2, 2025: “Donald Trump’s Rose Garden tariffs were historic in the most disruptive sense of the word. By raising tariffs on almost everything from almost every nation, he broke most of the trade promises America had ever made.” That epic level of promise-breaking will echo into the future of US diplomacy on all subjects.

The rise of the government statistical apparatus in the 20th century was heavily based on survey data, including surveys of households and businesses. But the willingness to respond to such surveys has been in decline. David S. Johnson, Maggie Meinhardt, and John Sabelhaus provide an overview of the problem in “Why did people stop responding to federal economic surveys? What can be done?” (Brookings Institution, April 14, 2026).

For a sense of the challenge, consider the falling survey response rates from households over time for key federal surveys. The CPS or Current Population Survey collects an array of data on households and income, along with employment and unemployment, poverty, and health insurance. Ten years ago the response rate was almost 90%; now it’s fallen almost to 60%. The CPI-Housing Survey is a monthly survey of 50,000 rental households for what they pay in rent–and the cost of shelter determines about 30% of the Consumer Price Index. The CE or Consumer Expenditure Survey collects detailed data on households and their detailed spending habits, sometimes through surveys and, for a certain group, through diaries.

Response rates are falling for business surveys as well. The CPI-C&S is a “Commodities and Services” survey that gathers price information from retail stores, service providers, and websites as an input to calculating the inflation rate. The CES or Current Employment Statistics survey collects data from about 700,000 workplaces on number of employees, hours worked, and earnings. The ECI or Employment Cost Index is based on data from the National Compensation Survey of businesses that collects data on wages, salaries, and benefits. The IPP or International Price Program survey collects data on prices of imports and exports.

The authors discuss the standard reasons for declining response rates: a proliferation of surveys that makes people and firms less willing to bother with any of them, concerns over privacy of results, and more. But what’s to be done?

One answer is to rely more heavily on “administrative” data: for example, instead of asking people in a survey to report what they earn, get the data from Social Security or income tax records. Instead of asking people in a survey what government benefits they receive, get the data from the government program making the payments. The good news about administrative data is that it’s often much more accurate than survey results. The bad news is that governmetn programs were not set up to be sources of data for social scientists. Their record-keeping isn’t organized in that way. Moreover, there are concerns about preserving privacy. There has been a considerable shift in this direction in economic research, although it’s still a work in progress.

As an additional step, one can imagine doing more to scrape the web for data on prices, job openings, and the like, although steps along these lines are already underway as well. I suppose one can also just ask an AI tool to survey the web and offer an estimate, but that approach is not yet ready for prime time.

But at the end of the day, some survey data isn’t easily replaceable. For example, the definition of “unemployment” is that a person is both out of work and also actively looking for a job. (If you aren’t actively looking, you are counted as “out of the labor force” rather than unemployed.) It’s not clear how to find out if people are looking for work unless you ask. Similarly, it is not always easy to get data on certain kinds of future expectations, like whether firms are planning to hire in the near future, without a survey tool.

The declining survey response rate is unlikely to reverse itself. But one can imagine a future where the surveys can be scaled back, when other approaches are available to gather the data, and can also become shorter so that they focus on the questions that can only be answered with survey data. The authors also note: Finally, although respondents to large-scale economic surveys like the CPS and CES participate voluntarily, other private and government surveys often use incentives—small gift cards or entries into a drawing for a larger prize—to compensate participants for their time.” Maybe the future of government data surveys will involve handing out lottery tickets, and publicizing the winners?

People in countries around the world would prefer to live in a a growing economy with opportunities for good jobs. Thus, it’s unsurprising that governments around the world would like to enact policies to deliver such outcomes. But industrial policy is hard: after all, if it was easy for governments to legislate prosperity, then there would be a lot less poverty and inequality in the world. Policies that favor one sector often involve costs both to the government and to other disfavored sectors. Ana Margarida Fernandes and Tristan Reed provide a menu of 15 industrial policy tools, and how and when they are more or less likely to work, in Industrial Policy for Development: Approaches in the 21st Century (World Bank, April 2026). Here is their menu:

You will notice that the possibilities are divided into first-rank and second-rank, where the second-rank choices specify that they make sense only when another option is not available. The authors make very clear that they are not suggesting every country should try everything on the menu, nor that these policies are certain to work. Instead, they are arguing that given details about the economy, government bandwidth, and fiscal situtation of the country, these policies are worth considering. But they offer many cautions:

Powerful interest groups often lobby for policies that benefit their constituents at disproportionate cost to the government. This can divert resources from more broadly valuable investments, such as education and health. Economists often describe this dynamic as government failure, offering it as an alternative explanation for why developing economies have struggled to build competitive industries.21 In such contexts, industrial policy can be inefficient or even harmful. For instance, subsidizing one industry may raise the costs of labor, capital, or raw materials for others. …

They offer a “decision sequence”:

(1) Keep emphasis on improving enabling institutions.

Despite the potential of well-designed industrial policies for development, nothing in this report suggests that they can be effective or efficient without enabling institutions. These institutions include accountable and capable implementing agencies that are insulated from politics and interest-group pressures, and strong economywide fundamentals: an educated and healthy workforce, energy and transportation infrastructure, and a sound macroeconomic framework. …

(2) Select low-cost public inputs not provided by the market.

Even with limited fiscal space and small local markets, countries with sufficient government bandwidth can still pursue an industrial strategy. The first choice should be public inputs that can be delivered at cost and are underprovided due to specific market failures, such as coordination failures (industrial parks), skills underinvestment (skills development), and information asymmetries (market access assistance and quality infrastructure). …

(3) Provide market incentives if fundamentals and public inputs are insufficient.

Countries should turn to market incentives as a last resort, as these are typically the most costly—either fiscally (production and innovation subsidies, consumer demand subsidies, and public procurement), for the broader economy (import tariffs, local content requirements, commodity export bans, export subsidies), or due to retaliation from trading partners. Moreover, these tools require careful monitoring. A notable exception is a technology transfer quid pro quo arrangement, when technology cannot be licensed, which incurs no fiscal cost.

(4) Be wary of macroeconomic interventions.

Competitive exchange rate devaluation is difficult to sustain over the long period of time needed to realize benefits and can trigger retaliation by other countries. More research is needed to understand whether and when general tax credits for research and development in private businesses translate into valuable inventions.

The authors also note:

Deciding which business activities are strategic is perhaps the most difficult and contested topic in industrial policy. As Nobel laureate Paul Krugman remarked about industrial policy in 1983: “While there is a valid case for targeting grounded in economic theory, the theoretical basis is too complex and ambiguous to be useful given the current state of knowledge”. Of course, the last four decades have seen significant progress in economic measurement, and recent years have seen a resurgence of interest in industrial policy among economists. Nonetheless, Krugman’s argument still largely holds.

The report offers lots to chew on about specific policies and the empirical record. Overall, where does it leave our thinking about industrial policy? My sense is that there’s relatively little controversy about the value of having better government institutions that focus on an educated and healthy workforce, strong macroeconomic funamentals, infrastructur, skills development, advice an networking on export promotion, and standard-setting, all in a way that is somewhat insulated from interest-group pressures. Indeed, my guess is that for a lot of people, these sorts of activities are just sound governmance, and they would reserve the term “industrial policy” for policies that provide fairly direct assistance to specific industries.

This narrower idea of industrial policy remains quite controversial. Advocates triumphantly cite cases where fairly direct assistance to specific industries (grants, cheap loans, tariff protection, government contracts, and the like) have worked out fairly well; cynics triumphantly cite cases where these forms of direct assistance do not seem to have helped even the targeted industry to develop, grow, innovate, and prosper, while still imposing substantial costs on the rest of the economy. Personally, I try to be open-minded but I lean to the cynical side.

For readers intersted in more on this subject, the world’s most dramatic economic success stories since the middle of the 20th century have all happened in nations of east Asia: Japan, Korea, China, as well as Thailand, Indonesia, Malaysia, Singapore, and Hong Kong. The Fall 2025 issue of the Journal of Economic Perspectives (where I work as Managing Editor)offers a two-paper mini-symposium on “The East Asian Tigers”:

• “Industrial Policy, Asian Miracle Style,” by Reda Cherif and Fuad Hasanov
• “The World Bank’s East Asian Miracle: Too Much a Product of Its Time?” by Nancy Birdsall

Predictions for how artificial intelligence technologies will affect future economic outcomes, for good or ill, are all quite dependent on the underlying assumptions. But it’s nonethelss interesting tow see what big international institutions are saying about the future. The recent report from the OECD, Foundations for Growth and Competitiveness 2026, is sweeping in its overview of the topic. Here, I will just focus on the subject of how AI technologies might affect growth.

Here’s an illustrative figure with estimates for seven high-income economies. The light-blue bars are AI-related annual productivity gains over the next decade with a slow-adoption scenario; the darker-blue bars are growth with a faster AI-adoption scenario; and the orange triangles are a medium-adoption scenario.

The OECD report offers some perspective on these growth rates: “For reference, the annual labour productivity growth, on average, has been 0.6% across G7 economies over the last ten years (2014-23), implying a growth-dividend from AI that is close to double the rate of recent productivity growth. In the most optimistic scenario, with fast adoption (as in mobile phones) and expanded AI capabilities, labour productivity gains are estimated at up to 1.3 percentage points, while in the most conservative scenario of slow adoption, they remain significant but at 0.2 to 0.4 points.”

To put it another way, say that AI raises US economic growth at the top end of these estimates, by 1.3% per year. The US GDP will be about $32 trillion this year. So a year from now, the US economy would be $416 billion bigger (that is, 1.3 percentage points) than it would otherwise have been. After 10 years, the GDP would be 13% bigger than it would otherwise have been (actually a bit more than that, because growth rates compound over time). This would be $4.16 trillion larger than the US economy would otherwise have been.

The empirical study underlying this figure, by Francesco Filippucci, Peter Gal, Katharina Laengle and Matthias Schief, is based on three components: “(1) micro-level productivity gains from AI at the task level, (2) the degree of exposure of tasks within sectors to AI, and (3) forecasts of future AI adoption across firms within each sector. To the extent possible, the quantification of these components relies on country-specific assumptions.” For perspective, that earlier paper also provides a range of estimates of the growth effects of AI. (The bars with stripes on top show the range from lower to higher estimates.)

But why are some economies, like the United States, getting a bigger boost from AI than others? Some answer are presented in “Mind the Gap: AI Adoption in Europe and the US,” by Alexander Bick, Adam Blandin, David J. Deming, Nicola Fuchs-Schündeln, and Jonas Jessen (Brookings Papers on Economic Activity, Spring 2026). As a big-picture overview, they point out that US investments in information and communications technologies have been substantially higher since at least the 1990s, and perhaps unsurprisingly, the US edge in output per hour worked has ben expanding since the mid-1990s, too.

But the real heart of their paper is survey data of workers, across countries, about their use of AI. For example, the US has a higher share of workers who report using generative AI tools.

In addition, looking just at the workers who use AI tools, the US workers spend a greater percentage of their work time using those tools.

The survey data from workers also lets the authors look at the personnel management policies across firms (building on questions asked by the World Management Survey). They find that companies and countries which score better on personnel management also tend to have higher AI use. In short, the US economy has a higher share of firms with highly-rated personnel management policies, and thus a higher share of AI users, and thus better prospects for a future AI-related boost in economic growth.

Tim Sablik of the Richmond Fed interviews “Ellen McGrattan: On measuring what businesses do, developing effective tax policy, and searching for answers beyond the lamppost” (Econ Focus: Federal Reserve Bank of Richmond, First/Second Quarter 2026). Here are a few of the comments that caught my eye:

How did McGrattan become interested in business cycles?

In grad school, I stumbled upon a paper by Finn Kydland and Ed Prescott that was game changing for me. Their revolutionary idea, which might sound obvious today, was that you should write down a theoretical model, simulate the data from it, and then match it up head to head with actual economic data. That wasn’t what economists were doing at the time. They would write down a theoretical model, understand its inner workings, and then put up a big wall between the model and the empirical work. Kydland and Prescott decided to take all the warts and pimples of an economic model, make predictions based on it, and go head to head with the data. That was the first time I had ever seen that done. I think it is the right approach.

It just so happens that what they were analyzing in that paper was business cycles. … Kydland and Prescott realized that adding the Fed wasn’t the answer, and they pointed to something else: Total factor productivity (TFP). Some might say that TFP is just a measure of our ignorance — it’s what we don’t understand. In some sense, I’ve been struggling with trying to look inside the black box of TFP all my career.

How studying total factor productivity led to intangible assets

My interest in business cycles partly stems from trying to measure what goes into TFP. … What is TFP? It’s what we don’t know, it’s the part that we need to fill in. It’s not just some magic dust that’s in the air. … If you buy, say, a computer, it has to be put on your balance sheet. But if you’re a dentist and you spend time building your patient list, that’s not put on any balance sheet. That patient list is the thing you sell when you retire or relocate, and that asset contributes to the value in the business, but we never see it until it gets sold or transferred somehow.

There are 40 million active businesses in the United States, and most have assets that we can’t see. Assets like customer bases or trademarks — until there’s a transaction, we can’t see them. You might have a good accounting system that you developed within your business, or you’re a chef and you have recipes, and we can only see those things if you trade them. But most ongoing businesses don’t list these assets on a balance sheet, so we never get to see them. And that’s why we need a theory to infer it. … It all ties back to work I was doing to measure movements in the economy over the business cycle, but now I would say the bigger issue is how to measure all activity in the U.S. economy.

McGrattan’s advice to students

 I always tell my students, ask a question first. Don’t read what other people have done. Decide on your own, especially when you’re writing your first paper. It doesn’t matter if you reinvent the wheel. If you’re thinking about things without having somebody else in your head, you’re going to come to a new creative idea. It’s fine after you’ve done something to compare yourself, because then you can really sharpen your results and make clear distinctions between what has been done in the past and what you did.

I really don’t like it when students are told to replicate findings from papers because that makes them too comfortable working with models that have already been analyzed to death. They learn an existing model and just make a small tweak to it. Before they do anything, I would rather they think of a good and, as yet, unanswered question and how they would go about answering it.

Often when I ask my students what they’re working on, they tell me they have some interesting data. That’s putting the cart before the horse. Start with an idea and then go down the path that may lead you to that data but may not. Don’t start with data and try to identify a question; start with a question and identify what you need to reliably answer that question.

Austin, Texas, is not small. The city itself has a population of about 1 million. The metro area (including the city itself) has a population of 2.5 million. About 10 years ago, Austin faced up to the issue of high and rising costs for rentals and for housing. It changes the rules for building in a way that expanded the housing stock by 30%. And rents fell. Liz Clifford, Seva Rodnyansky,  and Dennis Su of the Pew Foundation tell the story in “Austin’s Surge of New Housing Construction Drove Down Rents: Amid robust demand and a wave of policy reforms, Texas capital added 120,000 new homes from 2015 to 2024″ (March 18, 2026). They begin:

After decades of explosive growth, Austin, Texas, in the 2010s was a victim of its own success. Lured by high-tech jobs and the city’s hip reputation, too many people were competing for too few homes. From 2010 to 2019, rents in Austin increased nearly 93%—more than in any other major American city. And home sale prices increased 82%, more than in any other metro area in Texas.

But starting in 2015, Austin instituted an array of policy reforms aimed at encouraging the development of new housing, especially rentals. The city changed zoning regulations to allow construction of large apartment buildings, particularly near jobs and transit. In 2018, voters approved a $250 million bond measure to build and repair affordable housing. Permitting processes were reformed to speed development and reduce costs.

The efforts worked. From 2015 to 2024, Austin added 120,000 units to its housing stock—an increase of 30%, more than three times the overall rate of growth in the United States (9%).

Rents fell. In December 2021, Austin’s median rent was $1,546, near its highest level ever and 15% higher than the U.S. median ($1,346). By January 2026, Austin’s median rent had fallen to $1,296, 4% lower than that of the U.S. overall ($1,353). This decline occurred even though the city population grew by 18,000 residents from 2022 to 2024. In apartment buildings with 50 or more units, rents fell 7% from 2023 to 2024 alone—the steepest decline recorded in any large metropolitan area.

There seems to be a “folk economics” of housing. While many people believe in general that a higher supply will tend to drive down prices–and conversely, lower supply will tend to raise prices–these beliefs tend to go out the window when the subject of building more housing comes up. Instead, people begin to argue that building more housing only benefits real estate developer, or that any housing will only be for those with the highest income levels. Stan Oklobdzija, Christopher S. Elmendorf, and Clayton Nall describe this dynamic in “The Folk Economics of Housing” (Journal of Economic Perspectives, Summer 2025, where I work as Managing Editor):

[O]rdinary people simply do not believe that adding more housing to the regional stock would reduce housing prices. Across three original surveys of urban and suburban residents, only a minority of respondents say that a large, positive, regional housing supply shock would reduce prices or rents. These beliefs are weakly held and unstable (suggesting people have given the issue little thought), but respondents do have stable views about who is to blame for high housing prices: developers and landlords. Large, bipartisan supermajorities support price controls, demand subsidies, and restrictions on putative bad actors, policies which they believe would be more effective than supply liberalization for widespread affordability. 

People holding these beliefs might consider the Austin example. We aren’t talking here about a few more units of housing designated as “affordable,” or about subsidizing demand for housing and hoping the price comes down (!). If housing doesn’t seem affordable where you live, your local government should be thinking about what set of rules and zoning would make it possible to increase the housing supply by 20%, 30% or more over the medium-term of 5-10 years.

As the United States approaches its semiquincentennial (that is, half of 500 years) July 4, the McKinsey Global Institute offers a report that reads to me as a meditation on long-run US economic growth in “At 250, sustaining America’s competitive edge” (March 9, 2026). The US became the world’s largest economy in 1860, and has kept that lead since. In my reading, a major theme running through the report is the US leadership in originating and applying new technology. In that spirit, here are a few of the figures that caught my eye (but there’s much more in the report itself).

Here’s a proposed list of the top 100 technological innovations in products, systems, and technologies over the past 250 years. (I confess that I am a total sucker for mulling though lists like this.) The entries highlighted in dark blue were US-led; the entries in light blue were US-involved. In a broad sense, consider the dominance of US technological leadership since around 1900.

As US workers and companies used and developed these techologies for practical economic purposes, the McKinsey analysts suggest that it’s useful to think of US economic growth as proceeding through four stages of transformation. (Again, I’m a total sucker for whether these kinds of categories are the right break-points, or appropriately names, and so on, but I will leave you to your own meditations on these points and spare you my own.)

Combine this technological prowess with a legal and regulatory environment that supports a dynamic and evolving economy, along with investments in transportation and communications infrastructure, along with an enormous domestic market to sell into and (historically, although not so much at present) political support for agreements that let the US sell into international markets as wellm and one of the results is that the largest publicly traded firms in the world tend to based here in the United States. To put it another way, US citizens have the luxury of being worried about whether certain domestic firms are too successful and powerful, while people in other countries can only look on. It’s also useful to note that many of the biggest firms do turn over and evolve every few decades.

These and other factors discussed in the report (human capital, geography, investment, and others) add up to big-picture growth. The top panel shows the total size of the US economy, and points to the fact that it became the world’s largest economy in 1860. The bottom panel shows the per capita size of the US economy. Notice that when it comes to total size of an economy, China is #2. But when it comes to per capita size of an economy, China is still well behind the high-income countries of the world.

From a big picture view, the sustained growth and technologial leadership of the US economy is one of the remarkable world-historical facts of the last couple of centuries. We Americans are so close to it that we may fail to see it clearly. But now and again, in between the economic analysis and policy disputes, it’s worth a moment of wonder.

From the headlines, it looks as if momentum toward economic globalization has gone into reverse. But the underlying data is less clear. Steven A. Altman and Caroline R. Bastian lay out some evidence in the “DHL Global Connectedness Report 2026.” Some of their summary findings include:

The DHL Global Connectedness Index does not indicate a shift from international to domestic activity across trade, capital, information, and people flows. Global connectedness reached a record high in 2022 and has not changed appreciably through 2025. … U.S. tariff increases only modestly reduced forecast global trade growth. Other countries supported trade growth by not raising tariffs, and many negotiated new trade deals to secure access to alternative markets. … The world remains far from a split into disconnected geopolitical blocs. Only 4-6% of global goods trade, greenfield FDI, and cross-border M&A have shifted away from geopolitical rivals over the past decade. Trade flows shifted more toward neutral countries than to close allies, implying more ‘de-risking’ than ‘friendshoring’. … Prominent narratives about deglobalization are driven more by politics and public policy than by actual shifts in cross-border flows. While the risk of deglobalization has risen and the pattern of connectedness is shifting, the world overall remains as connected as ever.

Here are a few figures that caught my eye:

Global trade as a share of world GDP has been holding steady for a couple of decades now, with trade in goods sagging a bit and trade in services surging bit.

Global value chains–that is, goods where the stages of production happen in multiple countries–continue to lengthen.

The size of international capital investments, including foreign direct investment and portfolio stock investment have dropped a bit in these last few post-pandemic years, but remain near their all-time highs.

The report seeks to capture economically relevant information flows by looking at international collaboration in scientific articles, charges for use of foreign intellectual property, and international patent applications.

The report emphasizes that globalization still has plenty of room to expand:

[I]nternational flows of many types are close to record high levels relative to domestic activity. Nevertheless, we do not live in a “hyperglobalized” world. Most activity that could happen either within or across national borders is still domestic … The most recent available data show only 21% of all goods and services ending up in a different country from where they were produced. Companies buying, building, or reinvesting in foreign operations via FDI accounted for only 6% of gross fixed capital formation. Just 18% of traffic to online news websites came from abroad. And just shy of 4% of people lived outside of the countries where they were born.

Will globalization remain resilient, in the face of political pressures? Quite possibly. As the report points out, countries around the world outside the US clearly place a high value on the gains from trade; indeed, many of them are signing new trade agreements and keeping tariffs low with each other. The US economy is disengaging from direct trade with China, but in a multipolar world economy, this process often involves re-routing of trade, along with seeking out new opportunities, rather than a decline in the global total. Moreover, increases in trade have often been driven historically by new technologies–shipping, air freight, containerization, information flows. For example, the report cites a projection from the World Trade Organization that “AI could lift global trade in goods and services by 34–37% by 2040.” Love it or hate it, globalization doesn’t seem to be going away.

The GDP of North Korea was about $26 billion in 2024 (roughly equal to total sales for the Kraft Heinz Company), and with a population of about 26 million, that works out to a per capita GDP of (very) roughly $1,000 per year. However, that GDP estimate is not from any statistical release from the government of North Korea, which doesn’t do statistical releases. Instead, it’s from a report by the central bank of South Korea. Back in 2021, North Korea did send to the United Nations a report titled “Democratic People’s Republic of Korea Voluntary National Review on the Implementation of the 2030 Agenda. That report estimated a GDP of $33 billion in 2019–but without context on how that figure was derived.

How can one study an economy where government statistical data is unavailable or unreliable? Stephan Haggard, Kyoochul Kim, and Munseob Lee ask the question in “Studying economic black holes: Lessons from North Korea” (World Development, May 2026, 201: 107315). They discuss techniques of what they call “forensic economics.” “We focus on six sources of information: (i) data derived from remote sensing using satellites, (ii) information gleaned from periodic access by humanitarian agencies, (iii) `mirror statistics’ trade and aid data which relies on foreign partners rather than the government, (iv) information on prices collected surreptitiously, (v) information and data generated by refugee surveys and interviews, and (vi) creative use of official documents that unintentionally provide information that the government might actually seek to control.”

Perhaps the most famous illustration of satellite data for estimating GDP is from a 2014 NASA photo, which shows a view of “night lights” across South Korea, then the darkness over North Korea, and the night lights of China beyond. There has been active research seeking to infer GDP–and changes in GDP–from night lights and other satellite imagery. Haggard, Kim and Lee point out that South Korean central bank estimates of North Korea’s GDP are based on estimating output of key goods, like tons of steel or kilograms of rice, and then combining this with intelligence data. But of course, an estimate of GDP based on goods will miss the service sector. Apparently, South Korea’s estimates of North Korean GDP don’t show much increase in the last couple of decades, but estimates based on night lights do show a rise.

Estimates of agricultural output from satellite imagery can also be combined with data gathered from the UN Food and Agriculture Organization, when it is allowed to operate inside North Korea.

Estimates of North Korean trade patterns can be constructed from “mirror” statistics–that is, base on data from North Korea’s trading partners. China is by far the biggest trading partner for North Korea. The goods being traded can also offer insights into the underlying economy. For example, “the trend in the quality of Vietnam’s exported goods, measured in terms of human capital inputs, contrasts with that of North Korea. Vietnam’s exports have gradually shifted towards higher-quality goods, from rice and coffee to apparel and then electronics. In contrast, as diplomatic and military conflicts severed trade ties with Japan and South Korea in the late 2000s, North Korea expanded its trade with China, where economic growth fueled demand for fossil fuels. Consequently, North Korea’s anthracite coal exports surged, leading to a sharp increase in export volume but a decline in overall export quality. This suggests that North Korea’s heavy dependence on anthracite coal exports to China and reliance on other commodities could dampen long-run growth.”

Price data is considered a state secret in North Korea and “North Korea does not publish an official price index. Gathering pricing information from within North Korea requires the transfer of individuals or data across the country’s borders, at some risk to the participants given that the North Korean regime considers such data breaches as criminal and even capital offenses. However, complex information networks, relying on cell phones with the ability to connect to China, have allowed outside news organizations to put together consistent series of prices on basic commodities across multiple cities …”

Such data are sketchy, of course, but they pick up big patterns. This figure shows the price level in 2020 set equal to 100. As you can see, there was an extraordinary inflation between about 2010 and 2013, and then a milder inflation after 2020.

Surveys of refugees and defectors from North Korea are obviously not much help in estimating GDP, but they can offer detailed insight into particular industries–say, fishing–as well as day-to-day economic interactions.

Even official communications from the government of North Korea can offer insights, if cautiously digested. The authors explain:

North Korea is a particularly good example of an economic black hole because it publishes little if any usable economic data. However, authoritarian regimes do not simply stifle information; they use government-controlled media to communicate core messages to elites and to promote and legitimize their policies. The North Korean regime is no exception, and the array of such publications presents a potentially important field for the extraction of information, even if it requires careful reading; official news outlets; television broadcasts; speeches; and even academic journals reflect the government’s ideological stance and economic policies. In addition, the government publishes laws and regulations that pertain to economic activity and these have now been collated into accessible databases in English translation (North Korea Tech Lab, 202526). In general, the information available from these official North Korean sources has been underutilized by economists because it rarely provides the sorts of data that would be of direct interest to researchers. But recent advances in text mining through machine learning or so-called “text as data” approaches have enabled the conversion of qualitative data into quantitative data that opens up research opportunities.

North Korea is an extreme case of a black hole for economic data. But there are many countries around the world where estimates of GDP, or even basic estimates of population, can be quite uncertain. These tools of “forensic economics” have potentially much broader application.

Imagine that instead of dividing up US spending by program, you instead divide it up by what age group the money is spent upon. Kent Smetters takes this approach in “How Federal Spending is Distributed by Age” (Penn Wharton Budget Model, April 1, 2026). The results are perhaps unsurprising in general, but still eye-opening in their specifics.

The US government spent $7 trillion in 2025. In these calculations, $2.6 trillion of the spending can’t be easily assigned to a certain age group. As Smetters writes: “The largest items in the residual are net interest ($970 billion) and national defense ($917 billion). Transportation ($146 billion), natural resources and environment ($88 billion), administration of justice ($85 billion), community and regional development ($85 billion), and several other functions also remain in the all-ages category. These programs finance public goods, broad administration, place-based investment, or mixed services for households of all ages — they are not programs where recipient age is the central organizing concept.”

As one would expect, much of Social Security and Medicare can be assigned to the elderly, although these programs also cover a certain number of disabled persons who are not over the age of 65. Smetters writes: “Social Security directs $1.3 trillion to retirees, and Medicare sends $835 billion. Together they account for 80 percent of all age-assignable spending on older adults. But the retiree total extends beyond these two entitlements. Federal employee retirement benefits ($169 billion), housing assistance for older households, Medicaid long-term-care spending, and VA medical care all contribute, making the federal budget more retiree-focused than a Social Security–only lens would suggest.” Smetters finds that $2.7 trillion in federal spending goes to the elderly.

For federal spending on working-age adults, ” Medicaid directs $357 billion to working-age adults — more than to any other age bucket. Social Security also sends a substantial amount to working-age adults through disability and survivor pathways, contributing $216 billion. Beyond these two programs, Medicare for disabled beneficiaries under 65 ($159 billion), VA compensation and medical care ($202 billion combined), unemployment compensation ($40 billion), housing assistance ($44 billion), and Marketplace subsidies ($90 billion) all add materially.” Total federal spending directed to this group is $1.2 trillion.

Smetters combines federal spending on children and young adults under the age of 26 into a single group. “For people under age 26, Medicaid is the largest program, contributing $144 billion. SNAP follows at $44 billion, then child nutrition programs ($34 billion), Marketplace subsidies for younger enrollees ($29 billion), higher education support ($26 billion), elementary and secondary education ($24 billion), child care assistance ($21 billion), CHIP ($20 billion), TANF and family support ($18 billion), SSI for children ($15 billion), and foster care and adoption assistance ($11 billion). This is a different mix from the one that drives retiree spending: it is more fragmented and more dependent on means-tested programs.” Total spending here is $448 billion.

Put it together, and here’s a summary table. The punchline is in the final column. US spending on the elderly averaged $43,700 per person in 2025. In comparison, the average was $7,300 on the average working-age adult and $4,300 for the average child.

As I noted at the beginning, the overall pattern here is in unsurprising. Anyone who pays attention knows that federal spending on the elderly via Social Security and Medicare is huge. But of course, the elderly paid payroll taxes during their working lives to support others in these programs, and now it’s their turn to benefit from taxes paid by current workers. The amount directed to working-age people, as well as children, would be higher if one took into account federal benefits delivered through refundable tax credits like the earned income tax credit and the child tax credit, rather than focusing only on spending. Also, in the US system of government, a large portion of spending on children and young adults–like most of education spending and a substantial portion of Medicaid spending–happens at the state and local level, not the federal level.

Smetters, of course, is well aware of all of these issues. He is offering an intentionally blunt comparison to make a point. It’s pretty much always politically popular to offer higher benefits to the elderly, and to push back hard against proposals that seek to hold down the long-term rate of growth in these programs.

But about one-sixth of all US children live in families below the poverty line. If you look at families with income at or below twice the poverty line, which is often used as a measure of “near” poverty, then about two-fifths of US children live in such households. If you look at education as measured by the National Assessment of Educational Progress (often called the “Nation’s Report Card”), about 40% of fourth-graders test below the “basic” level of reading and 69% test below the “proficient” level.” In math, 24% of fourth-graders test below the “basic” level and 61% are below the “proficient” level. In short, the seeds of future inequality and poverty are happening with today’s children, right now. That per person federal spending gap between the elderly and the children/young adults group looks awfully large.