Conversable Economist

When I was first worrying about public policy issues, back in the late 1970s, discussions of fertility rates often invoked the 1968 best-seller, The Population Bomb, written by Paul Ehrlich. At that time, the global population was about 3.5 billion–roughly half its current level. But the book warned that it was already too late, that population was on the brink of overwhelming food supply and the environment, and that there would be mass famine around the globe in the 1970s. Even by the late 1970s, it was clear that the more apocalyptic predictions of the book were hyperbolic. But there was a multi-year famine in the early 1970s in the Sahel region of Africa (basically, a band across the continent reaching from parts of Senegal and Mauritania in the west to Sudan and Eritrea in the east).

I run into a fair number of people who seem to believe that the forecasts of the Population Bomb were only premature, not incorrect. My own sense is that when you predict impending overpoplation leading to mass famine, but then it doesn’t arrive in the next half-century even as population doubles, the odds are good that your analysis has overlooked some key ingredients. The current global food problem involves issues of both too little and too much: that is, about 9% of the world population is undernourished, but 40% of the world population (many of them in lw-income countries) are obese.

Moreover, the current concern seems not to be focused on overpopulation, but on the consequences of low fertility. The June 2025 issue of Finance & Development, published by the IMF, has several articles of interest about declining fertility rates.

David E. Bloom, Michael Kuhn, and Klaus Prettner provide an oveview of the arguments in “The Debate over Falling Fertility.” They write:

In 1950, the global total fertility rate was 5, meaning that the average woman in the world would have five children during her childbearing years, according to the United Nations Population Division. That was well above the 2.1 benchmark for long-term global population stability. Together with low and falling mortality, this drove global population to more than double over a half century, from 2.5 billion people in 1950 to 6.2 billion in 2000. A quarter of a century later, the world’s fertility rate stands at 2.24 and is projected to drop below 2.1 around 2050 (see Chart 1). This signals an eventual contraction of the world’s population, which the UN agency expects to top out at 10.3 billion in 2084. … Over the coming quarter century, 38 nations of more than 1 million people each will probably experience population declines, up from 21 in the past 25 years. Population loss in the coming quarter century will be largest in China with a drop of 155.8 million, Japan with 18 million, Russia with 7.9 million, Italy with 7.3 million, Ukraine with 7 million, and South Korea with 6.5 million …

For those concerned about overpopulation, this news must be concerning in the short run (global population will rise for the next few decades), but perhaps reassuring about the longer run (global population eventually set to decline.

Economists, of course, are proverbial for their every-rose-has-its-thorns mindset–that is, seeing the potential downside in all news. In that spirit, there are obvious concerns over lower fertility. For government finances, a much larger proportion of citizens will be elderly, thus relying on government pension and and health insurance benefits. Economic growth may slow down as well, with a greater share of the population either retired from working and/or starting new businesses.

There is occasional discussion of public policies to encourage families to have more children, but at least so far, countries that have adopted such policies have barely moved the needle of the overall fertility trends. My own sense is that the biggest pro-family policies include affordable family-style housing, high-quality schools and affordable higher-education, and a broader sense that of progress and economic growth.

Other articles in the same issue focus on a different adjustment: can people live longer in a healthy way, and perhaps also plan to work longer before retirement?

Andrew Scott and Peter Piot discuss “The Longevity Dividend,” by which they mean the ability of the healthy elderly to work more years. As they write: “The current health system is at risk of keeping us alive but not healthier for longer, at an ever-increasing cost to individuals, families, and society. In short, in the 20th century, we added years to life. In the 21st, we must add life to these extra years. This requires a shift toward chronic disease prevention and health maintenance, not just treating people when they become ill.” 

They describe a social and policy focus on what it would mean to have a society where the expectation for many people was that they would be in good health into their 80s. From an employment view, they add: “But good health alone is not enough to keep people engaged in employment for longer. We also need the kinds of age-friendly jobs older people prefer—with more flexible hours, fewer physical demands, and greater autonomy. By reducing the competition between younger and older workers, such jobs limit the career impact on the former.”

Similarly, Bertrand Gruss and Diaa Noreldin describe “Sustaining Growth in an Aging World.” They point out that people are remaining healthy later into life:

Data on individuals from 41 advanced and emerging market economies reveal that the recent cohorts of older people—those 50 and older—have better physical and cognitive capacities than earlier cohorts of the same age. When it comes to cognitive capacities, the 70s are indeed the new 50s: A person who was 70 in 2022 had the same cognitive health score as a 53-year-old in 2000. Older workers’ physical health—such as grip strength and lung capacity—has also improved. Better health means better labor market outcomes. Over a decade, the cumulative improvement in cognitive capacities experienced by someone aged 50 or over is associated with an increase of about 20 percentage points in the likelihood of remaining in the labor force. It’s also associated with an additional six hours worked per week and a 30 percent increase in earnings. All this could mitigate aging’s drag on growth.

It feels to me as if many people haven’t yet internalized what it means to have a reasonable expectation of living not only longer, but also healthier. I hear from (and about) people who just see it as a chance for a longer and more active retirement. But at least some older people could be enticed by more flexible labor market arrangements to keep a foot (or even just few toes) in the labor market for longer. I suspect that both they as individuals and society as a whole would benefit from that happening.

Back during the Great Recession of 2007-09, it became common for economists to talk about “financial plumbing” as part of the problem. The metaphor of pipes and drains and valves was meant to suggest that a relatively small blockage or capacity limitation in one part of the financial plumbing could lead to much bigger systemic effects. In other words, the financial plumbing might work just fine in ordinary day-to-day use, but if one part of the financial system came under stress, problems could back up unexpectedly.

With that general concept in mind, consider the total amount of US Treasury debt held by the public. Back in 2001, it was about $3.5 trillion. By 2009, it had doubled again to $7 trillion. By 2016, it had doubled again to $14 trillion. By 2024, it had doubled one more time to $28 trillion. The Congressional Budget Office forecasts suggest that by 2035, based on current law (that is, before the passage of this year’s budget and tax bills, which in their current form would increase the debt further), total US debt could nearly double one more time to $52 trillion.

So here’s the question: How confident should we be that the financial plumbing which handled the trading of US Treasury debt when the market was one-eighth of its current size, back in 2001, is equally capable of handling the much larger volume–especially when the market comes under stress? Darrell Duffie rings some warning bells in “How US Treasuries Can Remain the World’s Safe Haven” (Journal of Economic Perspectives, Spring 2025). (Full disclosure: I am the Managing Editor of JEP, and thus predisposed to find the articles of interest.)

If it seems far-fetched that the US Treasury market should come under stress, then it’s worth noting that it happened in March 2020. Duffie explains:

When the World Health Organization declared COVID-19 a global pandemic on March 12, 2020, … the dealers who make markets for Treasuries were unable to handle the flood of demands by investors around the world to buy their Treasury securities. Bond dealers were asked at the same time to buy enormous quantities of mortgage-
backed securities and corporate bonds, among other demands for liquidity. Total customer-to-dealer bond-market trade volumes suddenly jumped to over ten times their respective 2017–2022 sample medians (Duffie et al. 2023). The bond market reached the limits of its intermediation capacity and became effectively dysfunctional. Yields for Treasury securities lurched higher, while dealer-to-customer bid-offer spreads and dealer-to-dealer market depth worsened by factors of over ten (Duffie 2020). Among other steps to support the market, the Federal Reserve purchased almost $1 trillion dollars of Treasury securities from primary dealers in the first three weeks after March 12, freeing dealer balance-sheet space to handle more sales from customers. Weak market functionality persisted for several additional weeks (Duffie et al. 2023). Although liquidity in Treasury markets gradually returned to normal, many Treasuries investors presumably noticed that in the heart of the March 2020 crisis, they had not benefited from the safe-haven requirement of a liquid and deep market. Even before the COVID-19 crisis, the vaunted liquidity of the market for trading Treasuries had been showing cracks under stress.

Potential difficulties in the market for US Treasury bonds have large implications. As Duffie notes, many financial institutions and investors around the world view US Treasuries as their “safe” asset. Pretty much by definition, a safe asset holds its value and can be sold when desired. The ability of the US government to market its debt at favorable interest rates depends on this widespread perception. But if the market for Treasury debt can become illiquid in a crisis, as happened in March 2020, then Treasury debt is less safe than it previously appeared. The higher risk means that the US government would need to pay higher interest rates when it borrows.

As Duffie explains the plumbing in the market for Treasury debt, about $1 trillion is traded every day, and most of that flows through 25 firms that are designated as “primary dealers.” Essentially, this means that when there is a surge of sellers of Treasury debt, these primary dealers need to be financially able to act as immediate buyers–although of course they will be planning to re-sell most of that Treasury debt later. But the total amount of Treasury debt is rising fast, much faster than the financial size of the primary dealers. In 2007, before the Great Recession, the ratio of total Treasury debt to the assets of the primary dealers was less than 0.2; by 2023, the ratio was above 0.7. In short, the financial plumbing for the US Treasury market is running much closer to its capacity, and it has already gotten clogged once.

Of course, one way to make it easier for the primary dealers to guarantee that they will buy Treasury debt when needed would be to have less government borrowing and less Treasury debt. Now that we’ve all had a good giggle over the implausibility of that happening, what are the serious options? Ultimately, the goal might be to move beyond having the Treasury debt market flow through these 25 firms, and instead create an “all-to-all” market, more like the stock market, where buyers and sellers of Treasury debt can interact directly. But setting up such a market is nontrivial, and it still would raise the question of what happens in world financial markets if a wave of sellers of Treasury debt start driving down the price.

Duffie reviews a number of policy options, some of which are being implemented. You can read his article for details, but to give a sense of the possibilities:

• Require that trades for Treasury debt be carried out through a central clearinghouse, rather than as trades between two separate parties: “The clearinghouse offers a guarantee: if one of the original counterparties fails to perform at settlement, then the clearinghouse will complete the settlement.”
• “Regulators are slowly moving toward a plan for improving post-trade price transparency in the market for US Treasury securities by publishing trade price and quantities shortly after each trade (Liang 2022). Post-trade price transparency will likely improve competition and allocative efficiency. … The efficiency with which dealers are matched to trades will improve, likely expanding the intermediation capacity of the market. Eventually, greater post-trade price transparency will also speed up the emergence of all- to-all trade.”
• The Federal Reserve could set up arrangements to guarantee in advance that if/when US Treasury debt markets are melting down, they will extend short-term credit to key market players as needed. Experience has taught that when such backstop arrangements are known to be available in advance, they are less likely to become necessary!
• The US Treasury could buy back US Treasury debt issued in the distant past, which is harder to trade in the market, and replace it with newly-issued debt which is easier to trade in the market.
• Re-consider the specific bank supervision rules that try to make sure banks have sufficient capital to face crises, and make sure that these rules are not having the effect of discouraging banks from holding Treasury debt in a financial crisis situation.

Duffie says it bluntly: “The market for Treasury securities is simply growing too large to rely exclusively on dealers to intermediate investor trades.” Ultimately, the choice is whether financial regulators will proceed with all deliberate speed to implement the necessary changes before the next crisis hits the Treasury debt market, or whether the regulators will be improvising less-considered schemes when the next crisis hits the Treasury debt market.

Tim Sablik sets the stage for the conversation in “Carmen Reinhart: On twin financial and currency crises, the future of the dollar, and sovereign debt” (Econ Focus: Federal Reserve Bank of Richmond, Third Quarter 2025).

On the status of the US dollar as the global reserve currency:

When we talk about the dollar’s dominance, it’s important to first remember that central banks and investors are not buying greenbacks, they’re buying Treasuries. And it is the unmatched liquidity of the Treasury market that supports the role of the dollar. … When the euro came into being, for a while it looked like, while it might not replace the dollar, you could have a situation with dual reserve currencies. Before the global financial crisis, investors tended to view all European debt — whether it was French debt, German debt, Greek debt, or Irish debt — as close substitutes. Of course, the global financial crisis completely destroyed that perception. What it boils down to is that you have very fragmented debt markets in the eurozone that don’t offer the liquidity of the U.S. Treasury market. The euro is a unified currency, but there is no unification of the underlying assets that support the currency. Others have argued that the Chinese renminbi could be a contender to replace the dollar. I’ve never really entertained that possibility because, as Rudi Dornbusch used to say, people only go to a party if they think they can leave whenever they want to. China has capital controls, which directly impacts the liquidity of their debt market. How could you have as a reserve currency an underlying asset that in a time of need you can’t sell? 

On the challenges of high US government debt levels:

The recent surge in U.S. debt has outpaced many other advanced economies. Since the end of the pandemic, we’ve had ample opportunity — with a very tight labor market — to deliver more balanced budgets that would not continue to add to our debt, but we haven’t done so. Am I worried? Yes. Debt servicing has become more costly. Additionally, in the recent past when inflation and interest rates were low for long, volatility was suppressed. Now we have a combination that is much more difficult to manage: very high levels of debt, higher interest rates, and higher volatility.

The issue that I’ve always highlighted in my work is that there’s no silver bullet for dealing with high levels of public debt. Many countries might wish they could grow out of their debt, but that’s aspirational. Japan has been aspiring to grow out of its debt for decades. This is complicated by the fact that, as we discussed, growth is slower in periods of high debt burdens. That finding is based on long historic averages. If you look at Greece’s recovery from the global financial crisis, their per capita income in recent years was still below what it was before the crisis. So, there are no easy ways of delivering debt reduction. Growing out of the debt is unlikely if growth is slowing. Fiscal tightening is difficult. Inflation as a means of debt reduction is very unappealing. Debt restructuring, which is another term for defaulting, is also very unappealing.

On the social costs of a country defaulting on its debt:

The social costs of default have been overlooked in the economics literature. Typically, when one thinks of all the costs of default, there are the political costs, the fear of retaliation in terms of getting shut out of capital markets, and the economic costs. But related to those economic costs, you could think about the costs for households in terms of nutrition or health outcomes, for instance. And the research on those costs was a blank sheet. So, I wrote a paper with Juan Farah-Yacoub and Clemens Graf von Luckner, former students of mine, trying to quantify those costs. The results are pretty striking in terms of direction and duration. Life expectancy compares poorly versus the non-defaulters, and there is some increase in infant mortality. But the biggest effect that we see, apart from per capita GDP, is on poverty measures and things like caloric intake. So, the human toll of sovereign default is significant and long-lasting.

It is common in talking about 1980s to refer to it as a “lost decade” for economic growth, when many developing economies around the world were trapped in destructive patterns of debt and inflation. But William F. Maloney, Xavier Cirera, and Maria Marta Ferreyra make a stronger claim about growth in Latin America, which is that the causes and patternsof slow growth go back a century or more. They make the argument in the book, Reclaiming the Lost Century of Growth: Building Learning Economies in Latin America and the Caribbean (World Bank, 2025).

Consider some long-run patterns. This figure shows the countries of Latin America as a group with the orange line. The vertical axis shows Latin America and various high-income countries as a share of US income. Back in 1850, Latin America was at 30% of the US level; 170 years later, it’s about 25% of the US level. Meanwhile, countries like Sweden, Japan, Korea, Spain and Portugal have all experienced meaningful catch-up toward the US level.

This figure shows individual countries of Latin America, compared to Germany and France. As you can see, Argentina and Uruguay start out in 1850 above the levels for Germany and France, but have slumped since then.

The authors write: “Viewed through this lens, LAC did not lose two decades in the 1980s and 1990s; it lost the 20th century. …. [T]he LAC superstars hit a middle-income trap around 1900, while the poorer group grew well, but just not better than the frontier countries in other regions.”

The essential pattern here, argue Maloney, Cirera, and Ferreyra, is that the countries of Latin America have been systematically slower in assimilating and diffusing new technologies throughout their economies.

As one of many examples, the author point out that some countries in Latin Ameica had technological leadership of certain industries in the late 19th century: for example, Chile in copper mining and Mexico in mining of gold and silver. Indeed, the first technical school in the America’s, the Royal Mining College, was founded in Mexico in the late 18th century. But by the early 20th century, new techniques for mining were almost completely being developed outside Latin America. The authors write:

Further, mining had very different development impacts in other countries. Stanford University economic historian Gavin Wright cites the US experience with copper as an example of “how nations learn.” Exploiting copper gave impetus to schools of mining, for instance, at Columbia University and the University of California, Berkeley, which would later morph into major research universities at the frontier of metallurgy and chemistry, which in turn would lay the foundation for diversified industrialization. “The United States did more than passively live off the rents from these resources, but this unique resource base served as the foundation for an advanced national technology and applied science oriented toward this particular bundle of resources” (Wright 1987, 168). Similarly, Japan, perhaps contrary to its image as a manufacturing miracle, leveraged its position as a major copper producer in the same period into broader-based growth: high-tech conglomerates Fujitsu, Hitachi, and Sumitomo all began as copper mining companies (Maloney and Zambrano 2022).

Latin America lacks any comparable legacies. In fact, by 1900, mining across the continent had passed almost entirely into foreign hands, leaving the host nations with an acute feeling of dependency and a limited indigenous technical base upon which to diversify their economies. As late as 1952, Chileans had no capacity to monitor, let alone run, the giant foreign mines in the Norte Grande, and would not until 1965. By 1945, 96 percent of investment in the Mexican mining industry would be in foreign hands (Maloney and Zambrano 2022).

Multiply that example many times, and you have a picture of the economies of Latin America, for some decades now not catching up to the US standard of living, and not falling further behind. Stuck. The authors trace these patterns to a lack of engineering talent being trained in economies of Latin America; a lack of broad-based education that hindered the use of new technologies; a lack of entrepreneurs, and of capital sources to finance such entrepreneurs; and a lack of managers with the vision and resources to improve productivity in their own operations. In addition, Latin America from the middle of the 20th century often focused less on building a productivity advantage than on protecting domestic industries from the skills and technologies developed elsewhere. In the case of Mexico, the authors write:

Even adjusting for differences in economic structure from the average member country of the OECD, Mexico’s R&D intensity is the lowest, at about 70 percent of the level of Greece, Portugal, or Spain and 10 percent of the level of the highest countries (Austria, Finland, France, Sweden). Further, the share of R&D financed by the government ranges from 60 percent in Brazil to almost 80 percent in Mexico compared to 20 percent in China, Korea, and the United States, suggesting that LAC industries are investing proportionately far less. Mexico may represent the most extreme case of the innovation paradox. Geographically it sits immediately below the largest generator of new technology progress in the history of humankind yet somehow invests very little in the human capital or R&D needed to access it.

What is to be done? The discussion in this book focuses on incentives of existing firms to improve, the need for an expanded role for enterpreneurs, and the role of research institutes and universities. This all seems directionally correct to me. But I also find myself thinking that sustained and ongoing economic growth is also a process of sustained and ongoing social and economic change. For governments, firms, and people in countries that lag behind the technology and productivity frontier, catching up will require accepting and embracing an environment of change that will need to happen at a faster pace than change in countries that are technological leaders.

It seems plausible that the developing AI technologies will improve productivity. I have friends and family who tell me how it is already useful in their own everyday work, often for speeding up the process of producing first drafts: say, of a Powerpoint slide for display, or a recommendation for an internship, or the software programming to solve a problem. A common theme, it seems to me, is that the current form of AI tools can save considerable time at the earlier stages of a task, but at least for these kinds of tasks, the AI tools almost never get you close to a final version with the level of quality desired.

What about how AI might affect productivity growth for the economy as a whole? A group of OECD economists–Francesco Filippucci, Peter Gal, Katharina Laengle, Matthias Schief, and Filiz Unsal–offer an overview and discussion of existing estimates in “Opportunities and Risks of Artificial Intelligence for Productivity” (International Productivity Monitor, Spring 2025). Here’s their bottom line:

AI could contribute between 0.3 and 0.7 percentage points to annual aggregate
TFP growth in the United States over the next decade. The predicted impacts across different scenarios are highest in the United States, followed by the United Kingdom, Germany, Canada, France and Italy, and lowest in Japan. These figures indicate that Generative AI will likely be an important source of aggregate productivity growth over the next 10 years but also clarify that the expected gains from the current generation of AI technologies may not be extraordinary. For comparison, the latest technology driven boom linked to information and communication technologies (ICT) has been estimated to have contributed up to 1-1.5 percentage points to annual TFP growth in the United States during the decade starting in the mid-1990s …

This seems to be a mid-range forecast, compared to other published estimates. Here’s a figure showing the gains from use of AI in some specific studies for specific tasks. The vertical axis can be read as “%”– that is, AI tools improved customer service by 14%, improved coding by 56%, and so on.

How fast will AI tools be adopted? Here a figure showing roughly how long it took other systematically important technologies to be adopted by half of firms:

As these figures suggest, a range of factors can lead to higher or lower estimates for the productivity effects of AI. Here are a few of them:

• At what rate will AI tools be adopted across the economy?
• To what extent will AI speed up existing tasks for what share of current workers and jobs?
• To what extent will AI foster skill development for existing workers, allowing them to carry out additional tasks?
• Will making full use of AI tools require complementary investments in skills or physical capital that will take time and funding before they are implemented?
• To what extent will AI speed up research and innovation?
• Will the AI tools and the underlying technologies that allow use of those tools (like high-speed connectivity, computing power, and specialized computer chips) be made broadly available in a competitive market that tends to drive down the prices paid by users, or will they be sold in a less competitive environment with higher prices–so that they are used less frequently?
• To what extent will AI tools allow firms to exploit behavioral biases or malicious activities, in a way that results in lower overall gains for society?

The answers to these questions (and others) clearly open up an array of possibilities. As one final concern, the authors point out that “[h]istorically, sectors experiencing faster productivity growth have in fact tended to see decreases in their GDP shares (driven by declines in relative output prices and employment shares), thus reducing aggregate productivity growth …” Total productivity growth for the economy as a whole, of course, must include both sectors that make effective use of AI tools, along with sectors that don’t. The gains from AI in specific jobs and tasks is thus only one slice of the productivity picture.

I was recently reminded of the brief 201-word essay by E.B. White (1899-1985) on the meaning of democracy (which, I note, is subtly different than a straightforward definition of democracy). His most famous line is: “Democracy is the recurrent suspicion that more than half of the people are right more than half of the time.” Here is the whole thing, which appeared in the New Yorker on July 3, 1943. White wrote:

We received a letter from the Writers’ War Board the other day asking for a statement on “The Meaning of Democracy.” It presumably is our duty to comply with such a request, and it is certainly our pleasure.

Surely the Board knows what democracy is. It is the line that forms on the right. It is the don’t in don’t shove. It is the hole in the stuffed shirt through which the sawdust slowly trickles; it is the dent in the high hat. Democracy is the recurrent suspicion that more than half of the people are right more than half of the time. It is the feeling of privacy in the voting booths, the feeling of communion in the libraries, the feeling of vitality everywhere. Democracy is a letter to the editor. Democracy is the score at the beginning of the ninth. It is an idea which hasn’t been disproved yet, a song the words of which have not gone bad. It’s the mustard on the hot dog and the cream in the rationed coffee. Democracy is a request from a War Board, in the middle of a morning in the middle of a war, wanting to know what democracy is.

The US Declaration of Independence famously states: “We hold these truths to be self-evident, that all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness.” The Trinidad-born writer V.S. Naipaul, who won the Nobel Prize for Literature in 2001, offered a meditation on the broader sense of culture and human possibility that is contained within the idea of “pursuit of happiness” in a 1991 lecture (“The Universal Civilization,” New York Review of Books, January 31, 1991, quotation from p. 25). Naipaul wrote:

A later realization … has been the beauty of the idea of the pursuit of happiness. Familiar words, easy to take for granted; easy to misconstrue. This idea of the pursuit of happiness is at the heart of the attractiveness of the civilization to so many outside it or on its periphery. I find it marvelous to contemplate to what an extent, after two centuries, and after the terrible history of the earlier part of this century, the idea has come to a kind of fruition. It is an elastic idea; it fits all men. It implies a certain kind of society, a certain kind of awakened spirit. I don’t imagine my father’s parents would have been able to understand the idea. So much is contained in it: the idea of the individual, responsibility, choice, the life of the intellect, the idea of vocation and perfectibility and achievement. It is an immense human idea. It cannot be reduced to a fixed system. It cannot generate fanaticism. But it is known to exist; and because of that, other more rigid systems in the end blow away.

In the long run, a rising standard of living is always and everywhere based on productivity growth. Thus, Austan Goolsbee notedin a keynote address at the “Summit” conference held at the Stanford Institute of Economic Policy Research (SIEPR) in February (“Remarks on Productivity Growth and Monetary Policy,” February 28, 2025):

If you look at productivity growth, you’ll see that something weird and lovely has been happening over the past two years. Compared with the trend of the 11 years before Covid (figure 1), productivity growth since the end of 2022 has been notably faster. The slope of the line is steeper than the previous trend.

As Goolsbee notes, annual productivity numbers can jump around for a lot of reasons, and the period right after a pandemic might be especially prone to such jumps. Still, given the centrality of productivity growth and a couple of years of data, some speculation on possible causes doesn’t seem out of place. Goolsbee offers four possible explanations:

Economists have come up with four potential explanations. Three of those suggest this surge in productivity growth probably won’t continue.

The first explanation is that this is mostly just a reflection of the rise of work from home. … [I]f an increase in work from home drove the extra boost to productivity, that would be a one-time boost to the level of productivity, not a change to the overall growth rate going forward.

The second explanation is what economists call “labor reallocation and increased match quality.” Which kind of tells you why you should never ask for messaging advice from people with PhDs in economics. But this is just the idea that before Covid people were stuck in jobs they didn’t love and then the Great Resignation essentially let people rematch to do things that they are more motivated or better suited to do, and productivity went up. Even if you buy that as a driver, quit rates and other measures of job turnover are back to their pre-Covid levels, so the lovefest is probably done. This one, too, would be a one-time increase to the level of productivity, not a longer-lived change to the growth rate.

The third explanation is entrepreneurial dynamism: The number of startups each year was steady or falling for a long time, and it jumped at the start of Covid to a higher level and it hasn’t gone back down. But again, if new firms have higher productivity, this jump will show up as a one-time increase, not a sustained increase in the growth rate.

The fourth and final explanation is that this boom in productivity has been tech and AI driven. I realize that might have been where many of you first started, but note that economists are still skeptical—mainly because there hasn’t been enough adoption yet to explain why the economy-wide productivity growth rate would’ve increased this much. But here’s a key point, a key difference from the other three explanations: If this surge in productivity growth is the result of a new technology—whether that’s AI or something else—then history shows it is possible, that this surge is not just a one time bump. It could keep moving through the economy, industry by industry.

What evidence might distinguish between these theories? One potential angle is to look at productivity growth by industry, because some industries should be more heavily affected by certain explanations than others. Goolsbee comments:

Again, let’s remember if this surge were caused by work from home, labor reallocation, or more startups, we might expect to see broad-based, one-time-boost-type gains across many industries or concentrated in sectors with more work from home, et cetera. But that’s just not really what drove it. If you look at the industries experiencing the most significant productivity surge, seven or eight out of the top ten look tech or AI intensive … We’re talking about things like internet publishing, e-commerce, computer system design, renting intangible assets, motion picture and sound recording, and miscellaneous professional, scientific, and technical services.

Of course, this industry evidence doesn’t prove that AI is already the cause behind the productivity surge. It seems clear that one result of the pandemic was dramatically more widespread use of online services, including conferencing and related software tools, and that the tools themselves have gotten much better, too.

In the 2010s, it appeared that US health care expenditures as a share of GDP had peaked. But there group at the Centers for Medicare and Medicaid Services that continually carries out and updates these estimates and forecasts. Their most recent projections suggest that US health spending is about to start rising again as a share of GDP. Sean P. Keehan, Andrew J. Madison, John A. Poisal, Gigi A. Cuckler, Sheila D. Smith, Andrea M. Sisko,Jacqueline A. Fiore, and Kathryn E. Rennie present the estimates in”National Health Expenditure Projections, 2024–33: Despite Insurance Coverage Declines, Health To Grow As Share Of GDP” (Health Affairs, July 2025).

For perspective, health care spending as a share of US GDP had been rising steadily over the half-century prior to 2010: 5% of GDP in 1960, 6.9% of GDP in 1970, 8.9% of GDP in 1980, 12.1% of GDP in 1990, 13.3% of GDP in 2000, and 17.2% of GDP in 2010.

However, in the decade after 2010, US health care spending as a share of GDP seemed to stabilize: for example, it was 17.4% of GDP in 2018 and 17.5% of GDP in 2019. When the pandemic hit, and US health care expenditures leaped to 19.5% of GDP in 2020. But by 2022, health care expenditures had fallen back to 17.3% of GDP, similar to the prevailing level in the 2010s.

But looking ahead, the authors from CMS write: “Each year for the full 2024–33 projection period, national health care expenditure growth (averaging 5.8 percent) is expected to outpace that for the gross domestic product (GDP; averaging 4.3 percent) and to result in a health share of GDP that reaches 20.3 percent by 2033 (up from 17.6 percent in 2023).”

What’s behind that overall estimate? The authors explain:

During 2024–33, Medicare spending is projected to grow the most rapidly, at a rate of 7.8 percent annually … , mostly as a result of strong average enrollment growth compared with Medicaid and private health insurance as the last cohort of baby boomers ages into Medicare through 2029. Although yearly spending trends are projected to be more volatile for Medicaid than for other payers, annual Medicaid spending growth is projected to average 6.4 percent for the period 2024–33, which is about the same as the program’s average during the twenty-year period 2000–19. For private health insurance, high growth in utilization, along with notable changes in enrollment, are expected to moderate during 2024–33, with spending projected to increase by 5.2 percent, on average. … The out-of-pocket share of total health spending is projected to decline from 10.4 percent in 2023 to 9.1 percent by 2033.

Of course, projections should always be taken with a grain of salt. I’ll take up the question of how to moderate the rise in health care spending some other day. Here, I’ll just make three points. First, an increase in health care spending can be a good thing when it is accompanied by broad improvements in overall health statistics for the population, which does not especially seem to be the case for the United States. Second, when an ever-increasing share of GDP goes to producing health care, that money isn’t available for other socially desireable goals–including wage increases and public spending priorities. Third, every dollar spent on health care is a dollar in income to some participant in the health care system, all of who believe that their work benefits patients, and all of whom can be counted on to protest loudly any attempt to slow the growth rate of such spending.

Is it professionally worthwhile–not just as a form of light intellectual entertainment–for a modern economist to read articles and books written long ago? Do old article contain within them the possibility of live insights for modern economists? Matthew McCaffrey, Joseph T. Salerno, and Carmen-Elena Dorobat make the case for an affirmative answer in “The History of Economic Thought as a Living Laboratory” Cambridge Journal of Economics, March 2025, pp. 235-253). They write:

We argue that the history of thought can be conceived as a living laboratory of economic theorising. It is living in that it is a vital and valuable part of economics rather than a dead branch of it. It is a laboratory in that it functions as a proving ground in which theories from many different times and contexts can be examined, compared, critiqued, combined and developed. In other words, history of thought can be conceived as a method of doing economics rather than an isolated or niche field within it. …

For example, Axel Leijonhufvud similarly conceived of economics as a vast decision tree with many branches, each of which can be traced back to earlier choices by economists in a sequence potentially centuries long (Leijonhufvud, 2006). This is consistent with our own approach in that it views modern economics (the topmost branches of the tree) as embodying and reflecting a rich, living history of past choices and paths not taken (the lower branches or bits of tree trunk). Our approach expands some elements of the tree metaphor. In particular, our view is that HET [history of economic thought] is the tree itself rather than one of its dead branches or forgotten roots: there is no arbitrary point at which a branch becomes historical and therefore separate from the rest. Even the most recent branches exist synchronously with older ones. The decision tree of economics is dynamic, and old or seemingly withered branches can grow healthily again even after years of neglect.

The authors trace the “laboratory” metaphor back to Frank A. Fetter (1863-1949), who wrote:

Something of worth to present thought is, therefore, often to be gained by a restudy of past opinions, even though the first result may seem to be merely to expose their error. Showing that a thing cannot be done in a way that looks promising is often a service of laboratory research second only in value to showing how it can be done. The history of economic thought is the experimental laboratory of economics, or as near to that enviable agency of the physical sciences as social students are able to come.

They quote from Joseph Schumpeter’s (1883-1950) magisterial History of Economic Analyis, published posthumously in 1954:

[O]ur minds are apt to derive new inspiration from the study of the history of science. Some do so more than others, but there are probably few that do not derive from it any benefit at all. A man’s mind must be indeed sluggish if, standing back from the work of his time and beholding the wide mountain ranges of past thought, he does not experience a widening of his own horizon… But, besides inspiration every one of us may glean lessons from the history of his science that are useful, even though sometimes discouraging. We learn about both the futility and the fertility of controversies; about detours, wasted efforts, and blind alleys; about spells of arrested growth, about our dependence on chance, about how not to do things, about leeways to make up for. We learn to understand why we are as far as we actually are and also why we are not further. And we learn what succeeds and how and why.

Mark Blaug (1927-2011) took the laboratory metaphor a step further in the final pages of his Economic Theory in Retrospect, using it to emphasise the need for humility in economics, and for acknowledging the inescapable influence of the history of ideas:

One justification for the study of the history of economics, but of course only one, is that it provides a more extensive ‘laboratory’ in which to acquire methodological humility about the actual accomplishments of economics. Furthermore, it is a laboratory that every economist carries with him, whether he is aware of it or not. When someone claims to explain the determination of wages without bringing in marginal productivity, or to measure capital in its own physical units, or to demonstrate the benefits of the Invisible Hand by purely objective [i.e., without resorting to subjective value judgments] criteria, the average economist reacts almost instinctively but it is an instinct acquired by the lingering echoes of the history of the subject.

As McCaffrey, Salerno, and Dorobat argue: “In this version, the laboratory functions as a way of keeping economists grounded and avoiding mistaken claims of originality or inflated assertions of significance. It is also noteworthy that Blaug describes the history of thought as a ‘more extensive’ laboratory, implying the existence of a less extensive version. That version most likely refers to contemporary theorising without the benefit of history. In other words, Blaug too is suggesting that the history of ideas provides a richer and more expansive range of knowledge and ‘tools’ for economists to use. It is a crucial part of the economic record that gives us far more material to work with when determining the successes and failures of economic theory.”

I’ve read more history of economic thought than a number of modern economists; indeed, back in graduate school I even ploughed through Schumpeter’s thoroughly intimidating 1000-plus pages of the History of Economic Analysis. My own revealed choices suggest that I find it worthwhile to read the work of dead economists. But of course, my own preferences are not dispositive.

On one side, it seems to me that in questions of economics, inspiration can come from many sources. For some, inspiration arrives in the act of writing down and spelling out a mathematical model. For others, it arrives in the form of observations about the world that don’t seem well-explained by standard theory. For others, it arrives a part of the effort to answer a real-world policy question. Among the many potential sources of inspiration, it’s not hard to imagine that for some people, reading the efforts of earlier economists trying to come to grips with an issue might touch off a spark of insight.

But time is the consummate scarce resource. If an economist is looking for inspiration and insight, is it more likely to be discovered by spending the marginal hour reading articles by dead economists or by live ones? The answer probably depends on an idiosyncratic mix of reader and topic. But there is an intermediate choice, which is to be willing, now and again, to dip into the writings by those who specialize in history of economic thought. Their work offers a pre-chewed and partially digested version of past writings, which in more than a few cases has pushed me to dig deeper.

One of my friends used to say that “I’ll think about history of economic thought when I’m doomed to repeat it.” In that spirit, I’ll point out that many modern policy nostrums have actually been the subject of considerable debate. The arguments and evidence on how price controls work, or don’t, goes back centuries, as do the arguments about the effects of tariffs. Arguments about specific applications of price controls, like rent control, or about specific applications of tariffs, like favoring cerain domestic industries for purposes of national economic development, go back decades. If, say, rent control was a useful and obvious method of providing affordable housing across a large metro area or a country, one might expect to find lots of examples in the history of economic thought explaining how well it worked, and why. For me, history of economic thought often is useful in teaching humility, because it’s often true that very smart people have thought deeply about similar questions in the past. There’s a centuries-old saying to the effect: “If I see further, it’s because I stand on the shoulders of giants.” A neglect of the history of thought is an apparent belief that it saves time not to stand on the shoulders of giants.

As an editor, I’ll offer one other thought: If you haven’t actually read the article or book by a dead economist, be very cautious about repeating a quotation that you saw or heard someplace. A random quotation might seem like just a bit of color to enliven dreary prose, but if the quotation does not actually capture the original argument, then for those who have read the original, it marks you as unreliable.