The United States and Canada have a long history of friendly relations and, famously, share the longest undefended border in the world. Trade in goods and services has also linked the two countries with substantial economic benefits to both. By and large, trade flows reflect each country’s comparative advantage in producing goods and services. (We discuss the important role of comparative advantage in international trade in Microeconomics, Chapter 9 (Economics, Chapter 9 and Macroeconomics, Chapter 7).)
Image created by ChatGPT
The following figures show that in 2025, Canada was the leading market for U.S. exports and the second leading source of U.S. imports, behind only Mexico.
The two figures were prepared by ChatGPT using data from the U.S. Bureau of Economic Analysis.
Beyond trade in final goods and services, a number of U.S. and Canadian firms rely on capital goods and intermediate goods produced in the other country. For instance, in 2025, U.S. automobile manufacturers imported auto parts worth $19.5 billion from Canada. In other words, the supply chains of these firms rely on Canadian-produced parts.
Image created by ChatGPT
Economic relations between the United States and Canada have not always been smooth, however. In particular, the substantial increases in U.S. tariff rates in 1930 and during the second Trump administration resulted in sharp reactions from the Canadian government.
In 1930, Congress passed and President Herbert Hoover signed into law the Smoot-Hawley Tariff. In retaliation, Canadian Prime Minister William Lyon Mackenzie King and the Liberal Party significantly raised tariffs on U.S. imports. (We discussed the Smoot-Hawley Tariff in this blog post last year.) In the July 1930 Canadian elections, as the effects of the Great Depression began to be felt, Richard Bedford Bennett, the leader of the Conservative Party campaigned on using tariff increases to increase production and reduce unemployment. In a campaign speech, Bennett argued, “You have been taught to mock at tariffs and applaud free trade. Tell me, when did free trade fight for you? You say our tariffs are only for the manufacturers; I will make them fight for you as well. I will use them to blast a way into the markets that have been closed to you.”
Photo of Congressman Willis Hawley of Oregon and Senator Reed Smoot from the U.S. Library of Congress via the Wall Street Journal.
The Conservatives won an overwhelming victory in the 1930 election, and the Canadian Parliament passed legislation that raised Canadian tariff rates on U.S. imports to the highest levels in history. Bennett hoped that Canada could replace the decline in exports to the United States with an increase in exports to the United Kingdom. The following two figures, from an academic paper Tony published with his Lehigh colleague Judith MacDonald, indicate the unlikelihood of Bennett’s plan succeeding. For most of the twentieth century up to 1930 (with the exception of the World War I period), the share of Canadian exports that went to the United Kingdom had been declining, while the share that went to the United States had been increasing. In addition, in 1930, more than 60 percent of Canadian imports came from the United States as opposed to less than 20 percent coming from the United Kingdom.
For reasons of geography and the long-established trading relations between U.S. and Canadian firms, a major reorienting of Canada’s trade away from the United States and toward the United Kingdom wasn’t feasible. By 1935, near the end of his five-term, Bennett pivoted to attempting to negotiate a reciprocal trade agreement with the United States that would result in both countries reducing their tariffs on each other’s products. An agreement was reached in November 1935, but that was too late for Bennett who had been voted out of office in July.
The higher tariffs that the Trump administration has imposed on Canadian imports has placed Canadian Prime Minister Mark Carney in a situation similar to that Bennett faced in 1930. Like Bennett, Carney has responded to the higher tariffs by increasing tariffs on imports from the United States. And like Bennett, Carney has tried to find new markets outside of the United States for Canadian exports. According to an article in the Wall Street Journal:
“Carney has instructed his special envoy to Europe to scope out the most ambitious possibilities short of full membership in the [European Union] or its common market, according to people familiar with the matter. The details are still being sketched by technical working groups for what the prime minister has told his aides will be the reorienting of an economy and a society that for half a century has been dominated by the U.S.”
ChatGPT generated this image of the European Parliament building in Brussels, Belgium.
Carney’s plan of shifting Canadian exports from the United States to the European Union (EU) faces obstacles similar to those faced by Bennett as he attempted to substitute markets in the United Kingdom for markets in the United States. As the following figures show, in 2025, more than 70 percent of Canadian exports of goods went to the United States, while less than 6 percent went to the EU. Similarly, about 45 percent of the Canadian imports of goods were from the United States, while less than 12 percent were from the EU.
It may well be that Carney’s negotiations with officials in the EU are an attempt to push the United States into agreeing to reduce tariffs on Canadian imports. As a practical matter, though, it seems unlikely that Canada can reorient its trading relationships from the United Sates to the EU to any significant degree.
Today’s report from the Bureau of Labor Statistics (BLS) on the consumer price index (CPI) for August was eagerly awaited by economists and policy analysts. As we discuss in Macroeconomics, Chapter 15 (Economics, Chapter 25), monetary policy affects the economy with, in the words of Nobel Laureate Milton Friedman, “long and variable lags.” As a result, most economists agree that the Federal Reserve should not attempt to “fine tune” the economy by responding to each government release of macroeconomic data.
There are some instances, however, including the present, when the Fed’s policymaking Federal Open Market Committee (FOMC) appears to be uncertain as to whether a change in policy is needed. As a result, there was a widespread expectation that if today’s report indicated that inflation is slowing, the FOMC would likely leave its target for the federal funds rate unchanged at its meeting on Tuesday and Wednesday of next week. But if the report didn’t indicate that inflation is slowing, the committee would likely raise its target. The report gave few indications that inflation is slowing.
The following figure compares headline CPI inflation (the blue line) and core CPI inflation (the red line).
The headline inflation rate, which is measured by the percentage change in the CPI from the same month in the previous year, was 3.4 percent in August, the same as in July.
The core inflation rate,which excludes the prices of food and energy, was 2.4 percent in August, down from 2.5 in July.
Headline inflation was slightly higher, and core inflation was equal, to the forecasts of economists surveyed by FactSet. (Note that because of last year’s federal government shutdown, inflation data for October 2025 are not available.)
In the following figure, we look at the 1-month inflation rate for headline and core inflation—that is the annual inflation rate calculated by compounding the current month’s rate over an entire year. Calculated as the 1-month inflation rate, headline (the blue line) was 4.6 percent In August, up from 0.9 percent in July. Core inflation (the red line) was 3.5 percent in August, up from 2.6 percent in July.
The following figure illustrates the role played by energy prices in contributing to the large swings in the monthly inflation rate since the conflict in Iran began at the end of February. The red line shows the 1-month inflation rate in all energy prices included in the CPI. Inflation in energy prices, which had declined at an annual rate of 16.4 percent in July, increased at an annual rate of 28.3 percent in August. The blue line shows the 1-month inflation rate in gasoline prices, which had declined at an annual rate of 29.4 percent in July, increased at an annual rate of 58.3 percent in August.
There had been a fear that the rise in energy prices that began in March would pass through to increases in food prices, which are a key concern for many consumers. The following figure shows 1-month inflation in the CPI category “food at home” (the blue bar)—primarily food purchased at grocery stores—and in the category “food away from home” (the red bar)—primarily food purchased at restaurants. Grocery prices, which had declined at annual rate of 0.9 percent in July, increased at an annual rate of 0.4 percent in August. Food prices away from home increased 3.1 percent in August, down from 3.9 percent in July. To this point, increases in energy prices seem to have had some effect on grocery prices and restaurant prices, although the extent of the effect is unclear.
Fed Chair Kevin Warsh has indicated that he favors measures of the inflation rate that exclude particularly small or particularly large changes in the prices of some goods or services—so-called outliers. Median CPI, which is compiled monthly by economists at the Federal Reserve Bank of Cleveland, is calculated by ranking the price changes of every good or service in the index from the largest price change to the smallest price change, and then choosing the price change in the middle. The idea is to eliminate the effect on measured inflation of any short-lived events that cause the prices of some goods and services to be particularly high or particularly low. Economists at the Cleveland Fed have conducted research that shows that, in their words, “the median CPI provides a better signal of the underlying inflation trend than either the all-items CPI or the CPI excluding food and energy. The median CPI is even better at forecasting [personal consumption expenditures] PCE inflation in the near and longer term than the core PCE price index.”
Trimmed-mean inflation, also compiled by economists at the Cleveland Fed, excludes the highest 8 percent of price changes and the lowest 8 percent. The following figure shows 1-month trimmed mean (the blue line) and median (the red line) CPI inflation. Trimmed-mean inflation was 2.7 percent in August, unchanged from July. Median inflation was 2.1 percent in August, down from 3.1 percent in July. So these measures of inflation are both lower than the conventional headline and core CPI inflation measures, although as the figure shows, both measures are volatile.
Note that the Fed uses the 12-month change in the personal consumption expenditures (PCE) price index, not the change in the CPI, when gauging whether it is hitting its 2 percent annual inflation target. Historically, PCE inflation has been about 0.4 percentage points to 0.5 percentage points lower than CPI inflation. The Bureau of Economic Analysis (BEA) won’t release its estimate of August PCE inflation until September 30, after the next FOMC meeting.
Today’s report showing that inflation remains persistently above the Fed’s 2 percent annual target, following last week’s jobs report showing an unexpectedly large increase in employment, has likely raised the chance that Federal Reserve policymakers will increase their target range for the federal funds rate from the current 3.50 percent to 3.75 percent by o.25 percentage points (or 25 basis points) at the next meeting of the FOMC on September 15–16. Trading in the federal funds futures market this afternoon indicates that investors assign a 86.5 percent probability to the FOMC raising its target range at that meeting, which is up from a 72.4 probability yesterday. Trading indicated that investors assign a 74.5 percent probability to the committee increasing its target range by at least 50 basis points by the end of the year, up from 64.6 percent yesterday and from 44.7 percent one week ago.
In Microeconomics, Chapter 17, we discuss the tax system, including principles and goals policymakers can use to evaluate tax proposals. In this column, which first appeared on the website of the American Enterprise Institute, Glenn discusses the advantages of the federal government switching to using a cashflow tax, rather than the current personal and corporate income taxes.
The Case for a Cashflow Tax
In the context of the US economy, a cashflow tax would offer far-reaching benefits in promoting economic growth, taxing rents, and raising incremental revenue efficiently. Moreover, well-known concerns about complexity and fairness are easily addressed.
After the U.S. midterm election this November, Congress’s economic-policy focus will likely turn to growth, AI’s impact on the economy, and deficit concerns. Lawmakers therefore should recognize that one policy tool can help to address all three: a cashflow tax.
Attention to these issues is warranted. Because higher productivity is what boosts incomes, investment and growth remain the core drivers of rising living standards. Yet concerns over labor’s shrinking share of GDP relative to capital—a problem that could be magnified by developments in generative AI—have ignited debates about profit and wealth taxation. At the same time, growing alarm over the unsustainability of the U.S. fiscal position puts a premium on efficient ways of raising additional revenue.
That is where the cashflow tax comes in. Many economists (including me) have long championed the economic benefits of moving the tax system further toward a consumption tax. Relative to the current income tax, a broad-based consumption tax would raise saving and investment, leading to higher productivity and incomes.
Though the term “consumption tax” may call to mind a European credit-invoice value-added tax (VAT), the United States could implement such a policy in a simpler fashion by targeting accounts currently maintained for taxation. After all, a consumption tax at a given rate is arithmetically identical to the combination of a wage tax and a business cashflow tax at the same rate. As a tax on a firm’s revenue minus expenses, a cashflow tax would allow businesses to expense investment immediately, boosting outlays.
Such a shift would expand on reforms enacted in 2017 and 2025. It would also disallow nonfinancial companies from making interest deductions, because in contrast to an income tax, a cashflow tax treats debt and equity the same. It therefore eliminates an important tax incentive for firms to allow themselves to become leveraged. Finally, a cashflow tax is much simpler than a corporate-income tax, which requires complex depreciation schedules.
Cashflow taxation shifts the tax burden toward high rates of profit, which is useful in an environment of rising profit concentration in firms. This benefit arises because the cashflow tax removes taxes on what economists call the “normal return” on investment (or the cost of capital). Companies would instead be taxed only on profits above this amount, reflecting economic rents. Moreover, given that most large individual fortunes reflect economic rents in business ownership, a cashflow tax would be more effective than a wealth tax, which suffers from many complications relating to measurement, liquidity, and incentives.
Because a cashflow tax creates fewer distortions of saving and investment than the income tax, it is also a more efficient instrument for raising incremental revenue in any future fiscal consolidation. While the bulk of fiscal adjustment will require reductions in the growth of federal spending, higher revenue would almost surely be part of any politically viable package.
Moreover, additional revenue could be raised if Congress incorporated a border adjustment in the new cashflow tax. Doing so would deny companies a tax deduction for expenses abroad, while exempting U.S. exports from taxation. Other countries already use border adjustments in their VATs on consumption, and America uses a similar mechanism in state and local retail taxes. If you buy a kitchen appliance in New York, you pay New York sales tax even if it was made in Ohio. The sales tax applies only where the good is sold, not where it originates. Because the U.S. imports more than it exports, the border adjustment would raise revenue, remove tax incentives for U.S. firms to locate activities abroad, and strengthen the incentives for non-U.S. firms to locate activities in America.
To be sure, while a cashflow tax promises to promote economic growth, tax high profits, and raise incremental revenue efficiently, it does raise concerns about complexity and tax fairness. But these issues can be straightforwardly addressed.
Consider the possible concerns over tax complexity. The cashflow tax would use basic company accounts already used in the income tax, avoiding any new fundamental tax design or introduction of a European-styled VAT. It also would offer simplification within the income tax code by substituting permanent expensing of capital goods for more complicated and less generous depreciation allowances.
Yes, the combination of a wage tax and a cashflow tax at the same rate replicates a broad-based consumption tax, but without the progressivity of the current income tax, raising potential concerns about fairness. But this issue is also easily tackled. A graduated tax structure could be introduced along the lines of the “X-tax” that Princeton University economist David Bradford developed 40 years ago. Here, the individual wage tax would have graduated rates with a credit for low-wage individuals, and the business cashflow tax rate would be set as the highest individual tax rate. Or, as Treasury Secretary Nicholas F. Brady suggested in 1992, an alternative version would tax all income, including capital income, for very high-income individual taxpayers.
With elections looming, economic policies to promote growth, address structural changes in income and wealth inequality, and tackle ballooning budget deficits will be intensely debated. Building on recent tax changes to implement a cashflow tax offers the most promising path forward on all three issues.
A recent article in the Wall Street Journal discussed the surprising fact that some colleges are sending letters of acceptance to students who haven’t actually applied for admission:
“Hundreds of colleges are sending students letters of admission—without even requiring an application. … Known as ‘direct admissions,’ this expedited process is free and omits required essays, questions about extracurriculars and mandated standardized tests.”
The following figure from the article shows the increase in the number of colleges among the 1,100 colleges that accept the Common Application (or Common App) that use direct admissions.
The rise in the use of direct admissions reflects a decline in students’ demand for admission to these schools. Part of the reason for this decline in demand is the falling number of people in the United States who are in the prime college attending ages of 18 to 24. The following figure shows projections from the Census Bureau of the number of U.S. residents in this age group from the present to the year 2100. The numbers on the vertical axis are thousands of persons. From 2022 to 2026, the number of people in this age group declined by about 1 million. The number is projected to have declined by another 2 million in 2040.
Another factor that may be affecting the demand for college admissions is stagnation in the college wage premium, which is the amount by which wages earned by college graduates exceed wages earned by high school graduates. The following figure from a publication of the Federal Reserve Bank of Minneapolis shows values for the college wage premium from 1961 to 2023. The figure uses data from a working paper by economists at the Federal Reserve Bank of San Francisco that adjusts the college wage premium to take into account several factors, including differences in the ages of high school and college graduates.
The college wage premium has fluctuated, but from 1980 to 2000 it was generally increasing. Since 2000, however, the premium has stagnated. Several explanations have been offered for this stagnation. Lisa Camner McKay of the Minneapolis Fed notes that the relative supply of workers with college degrees has been increasing: “In 2000, workers with a bachelor’s degree or higher were 31 percent of the civilian labor force. In January 2025, they were 45 percent.”
The labor market demand for college graduates may also have declined relative to the demand for high school graduates. The following figure, based on data in the working paper from the San Francisco Fed referred to earlier, shows the ratio of public job postings that require applicants to have a college degree relative to job posting that don’t require a college degree. The ratio has steadily declined since 2010.
We’ve identified two factors that may account for a decline in the demand for a college degree that’s led some colleges to rely on direct admissions to recruit students. Media stories have also noted that some smaller colleges have been forced to close in recent years as they were unable to recruit enough students to cover their costs. These closings have reduced the supply of college degrees. However, only about 46 traditional nonprofit private colleges closed between 2023 and 2025. While these closures have been a hardship for the students, faculty, and administrators involved, they have been a very small fraction of the more than 3,000 public and private colleges in the United States. But some observers have forecast that closures of small private colleges may sharply increase in the coming years. For example, an article in the Wall Street Journal cited a study by Huron Consulting that found that 442 of the 1,700 private nonprofit colleges have experienced shrinking enrollments and are at risk of closing at some point in the next 10 years.
How might declines in the demand for and supply of college degrees affect the tuition that students will pay in the future? First, it’s worth noting that, corrected for the effects of inflation, college tuition has not increased significantly in recent years. The following figure, using data from the College Board, shows that, when measured in 2025 dollars, college tuition at public and private colleges has been roughly flat over the past 10 years, particularly if we look at net tuition charged, which subtract grants the colleges have awarded to students from the colleges’ published tuition amounts.
We can use the model of demand and supply to analyze how tuition might change in the future. In Microeconomics, Chapter 3, Section 3, we show that whether the price in a market rises over time depends on the direction in which demand and supply curves shift and on the relative magnitudes of the shifts. In this case, our discussion indicates that both the demand for college degrees and the supply of college degrees are likely to continue shifting to the left. Whether tuition rises or falls depends on the magnitude of the shifts. If the shift in demand is greater than the shift in supply, tuition will fall. If the shift in supply is greater than the shift in demand, tuition will rise. The following figure illustrates the situation in which the demand for college degrees shifts by more than the supply of college degrees, causing tuition to fall.
Image created by ChatGPT, as are the other images in this post.
Technological breakthroughs are often embodied in machinery, equipment, or, as in the case of artificial intelligence (AI), software. For example, the Industrial Revolution of the late 1700s, which involved mass production of low-priced cotton textiles, was made possible by the development in England of large cotton spinning machines that were often powered by steam engines.
Other technological breakthroughs are disembodied because they involve reorganizing production rather than employing new machinery or software. For example, the just-in-time (JIT) system that Toyota developed beginning in the 1940s, lowered the costs of assembling automobiles by, among other things, having parts arrive at a factory just when needed.
Across U.S. history, there have been a number of technological revolutions that required massive capital investments. For instance, the transportation revolution of the early 1800s that significantly reduced the costs of shipping freight, relied on substantial investments in canals. Later in the 1800s, transportation costs were reduced further as firms invested in the largest network of railroads in the world. Electrification of businesses and private homes in the late 1800s and early 1900s required the building of power plants and transmission lines. In the late 1990s, many firms competed to build networks of fiber-optic cable to carry the rapid increase in digital traffic resulting from the development of the internet.
Each of these technological revolutions succeeded in providing consumers with new or lower-cost goods or services. But building the physical capital that embodied these technologies could result in substantial losses or bankruptcy. For example, in the 1830s, several states, including Pennsylvania, Illinois, and Michigan, defaulted on the bonds they had issued to finance canal instruction.
Similarly, in the 1890s, some of the largest private railroad companies in the United States, including the Union Pacific, the Northern Pacific, and the Atchison, Topeka & Santa Fe, went bankrupt. In the early 2000, investors realized that internet commerce wouldn’t expand as rapidly as they had expected. As a result, several prominent internet firms, such as Pets.com and eToys.com, saw their stock prices collapse in what was called the dot com crash, and eventually declared bankruptcy. Several firms that had invested in building large fiber-optic networks, including Worldcom and Global Crossing, also failed.
Why did the companies making large capital investments in new technologies sometimes fail even though the technologies themselves proved to be revolutionary? The answer in part is that forecasting accurately the demand for a new product can be difficult. For example, ultimately the level of internet traffic was more than great enough to justify the investment in fiber-optic networks, but the increase came too late for some companies to avoid bankruptcy. Competition among firms to take advantage of a new technology can lead to overbuilding, as when railroad companies built more miles of track than were needed to meet the demand to transport freight and passengers.
A number of economists and investment analysts have noted that the very heavy investment in data centers used to power AI models resembles the large capital investments firms made during earlier technological revolutions. As the following figure shows, beginning in early 2023, shortly after the public release of ChatGPT, the first widely used AI chatbot, investment in information processing equipment and software as a percentage of all nonresidential fixed investment has increased dramatically.
Do the firms building data centers run the risk of overestimating demand for AI and overbuilding data centers? More generally, how does the build-out of AI data centers compare with the earlier build-outs of railroads and fiber-optic networks? Who better to answer this question than ChatGPT!
The following (very long!) answer was ChatGPT’s response to the following prompt: “In what ways is the current build-out of AI data centers similar to the build-out of the railroad network in the United States in the late nineteenth century and the build-out of the fiber-optic network in the United States in the 1990s and in what ways is it different?” (A quick summary is given in the brief “Bottom Line” section at the very end.)
AI Data Centers, Railroads, and Fiber-Optic Networks
The best way to frame the comparison is this:
The railroad analogy is most useful for understanding the physical scale, bottlenecks, public subsidies, and geographic consequences of the AI buildout. The fiber-optic analogy is most useful for understanding its financial risks, uncertain demand, rapid technological improvement, and potential for overcapacity.
AI data centers nevertheless differ importantly from both. They are not simply links in a transportation or communications network. They are more like factories that produce computing services, and the extremely expensive machinery inside them becomes obsolete much faster than either railroad rights-of-way or buried fiber.
The scale is already extraordinary. The International Energy Agency estimates that the largest technology companies spent more than $400 billion on capital investment in 2025 and expects that figure to rise about 75 percent in 2026. U.S. data centers consumed approximately 176 terawatt-hours of electricity in 2023—4.4 percent of U.S. electricity use—and Berkeley Lab projects consumption of 325–580 terawatt-hours by 2028.
Ways the three buildouts are similar
1. All three are investments in enabling infrastructure
Railroads did much more than generate revenue from tickets and freight charges. They brought previously isolated regions into a national market, encouraged settlement, changed the location of factories and cities, and made complementary investments in farms, mines, warehouses, and factories more valuable. The United States added roughly 170,000 miles of railroad track between 1871 and 1900.
Fiber-optic builders similarly expected abundant bandwidth to generate applications that could not yet be foreseen. More bandwidth would encourage new internet services, which would increase bandwidth demand, which would justify still more fiber.
The AI argument is nearly identical: companies are building computing capacity before all its profitable applications have been discovered. The value of the data centers will depend partly on complementary developments—better models, proprietary data, new software, organizational changes, and the willingness of companies and consumers to pay for AI services.
In all three cases, therefore, infrastructure and applications are jointly determined. Waiting until demand is plainly visible can mean waiting too long, because the infrastructure itself helps create the demand.
2. High fixed costs encourage companies to build ahead of demand
A railroad requires land, grading, bridges, tunnels, stations, track, and rolling stock before it carries its first paying customer. A fiber network requires rights-of-way, trenches, cable, switching equipment, and connections before it can sell substantial bandwidth.
AI data centers likewise require sites, grid connections, substations, transformers, cooling systems, generators, networking equipment, and large orders for advanced chips. Many of those expenditures are sunk: once a company has built a substation or a specialized data-center shell in a particular location, much of the investment cannot readily be recovered.
Companies also face a strategic incentive to build early. An individual hyperscaler may reasonably conclude that the loss from having insufficient computing capacity—lost customers, inferior models, or technological dependence on a competitor—would be greater than the loss from temporarily having too much. But when every major company reasons this way, the industry can collectively build more capacity than it needs. The IEA reports increasingly severe bottlenecks involving electricity, grid connections, chips, memory, transformers, and capital, even while acknowledging that many announced projects will never be completed.
3. Each boom rests on highly uncertain projections of future demand
Railroad promoters projected the future settlement, agricultural production, and mineral output of regions that were sometimes scarcely populated. Eventually, railroad projects began to outpace the demand for their capacity, reducing returns and contributing to widespread defaults.
Fiber builders extrapolated the extraordinary early growth of internet traffic. They assumed that new applications and fiber connections to homes and businesses would arrive rapidly enough to absorb the new long-distance capacity. Those assumptions proved too optimistic, at least in the short run.
AI companies are making similarly difficult forecasts. No one knows with much confidence:
how much computing future models will require;
how quickly businesses will reorganize around AI;
how much customers will pay for AI services;
whether most AI revenue will accrue to model developers, cloud companies, application providers, or users; or
how much improved chips and algorithms will reduce the computing required for a given task.
The uncertainty does not mean that the investment is irrational. It means that large forecast errors—in both directions—are likely.
4. Rapid technological progress can itself produce overcapacity
This was especially important during the fiber boom. In 1996, one fiber strand could carry approximately 2.5 gigabits per second. By 2000, improved multiplexing equipment could raise the capacity of the same strand to approximately 100 gigabits per second. Thus, companies were not merely laying more fiber; technology was multiplying the capacity of fiber already in the ground. Communications-equipment investment rose from about $62 billion at an annual rate in early 1996 to more than $135 billion in late 2000, before falling sharply.
AI faces an analogous risk. The cost of running a model with roughly GPT-3.5-level performance fell more than 280-fold between November 2022 and October 2024, according to Stanford’s AI Index. Hardware costs were declining by about 30 percent annually and energy efficiency was improving by approximately 40 percent annually.
Consequently, today’s estimates of the number of data centers needed in 2030 may prove too high if chips, models, and software become much more efficient. That is the closest parallel to the fiber glut.
But there is an important countervailing effect: cheaper computation can generate much more use. The IEA estimates that energy consumption per basic AI task has recently fallen by at least an order of magnitude annually, yet more demanding reasoning, video, and agentic applications are expanding so quickly that total data-center electricity consumption is still projected to roughly double between 2025 and 2030. Efficiency may therefore lower the cost of AI without lowering total demand for computing or electricity.
5. A private investment failure can leave a valuable social legacy
Many railroad investors lost money, and many railroad companies entered receivership, even though the resulting network transformed the American economy. The Panic of 1873 began in substantial part because railroad construction had outrun demand and heavily indebted companies could not refinance, but the tracks themselves did not cease to be useful.
The fiber experience was similar. Long-distance fiber companies failed, telecommunications shares collapsed, and equipment investment fell, but the underlying fiber remained available to carry later increases in internet traffic. The glut helped drive down the price of communications capacity and made subsequent internet businesses less expensive to operate. The Richmond Fed concluded that excessive investment and rational responses to genuinely revolutionary—but highly uncertain—technology were not mutually exclusive explanations.
AI could follow the same pattern. A future decline in data-center rents or computing prices might be bad for data-center owners and chip suppliers but good for application developers and users. AI could be economically transformative while a substantial fraction of current AI investment earns a disappointing return.
Important differences
1. A data center is not a network in quite the same sense
A railroad line moves something between two locations. A fiber-optic cable transmits information between locations. Their usefulness depends strongly on connections to the rest of the network.
A data center primarily produces computation. It resembles an electricity-generating plant, steel mill, or semiconductor fabrication plant more than a railroad line. It needs fiber connections to communicate with users and other data centers, but its principal economic output is processing rather than transportation.
This matters because railroad and fiber networks have especially strong physical network effects: a route becomes more valuable when it connects to additional routes. AI has powerful scale economies and ecosystem effects, but much of that advantage resides in models, data, software, cloud platforms, and proprietary chip architectures rather than merely in connecting one data-center building to another.
2. Technological progress affects the installed assets differently
With fiber, new electronics could greatly increase the capacity of fiber that was already buried. Technological improvement therefore often enhanced the usefulness of the old physical asset.
With AI, a new generation of accelerators may make the previous generation significantly less economical. The site, building, electrical connection, cooling system, and fiber links can remain useful, but the chips and servers may have to be replaced.
Meta’s 2025 annual report estimated useful lives of five to five-and-a-half years for servers and network assets, compared with 25–30 years for its buildings. Even five years may overstate the competitive life of some frontier AI equipment if newer chips provide much better performance per dollar or per watt.
This makes unused AI capacity more perishable than “dark fiber.” A buried fiber strand can sit unused for years and later be equipped with improved electronics. A warehouse full of unused accelerators loses value simply as better chips arrive.
3. AI data centers have unusually large continuing resource costs
After fiber has been installed, transmitting another unit of information has a very low marginal cost. Railroads have significant continuing costs—labor, fuel, maintenance, and rolling stock—but the track itself does not require a massive continuous energy input merely to remain economically relevant.
AI computation consumes large amounts of electricity whenever it is used and requires cooling, maintenance, networking, and periodic hardware replacement. The IEA expects global data-center electricity consumption to rise from approximately 485 terawatt-hours in 2025 to about 950 terawatt-hours in 2030. AI-related servers are expected to account for much of the increase.
Thus, an overbuilt AI center can continue generating costs even after the initial construction bill has been paid. Its economics depend not merely on filling the building but on earning enough revenue to cover electricity, maintenance, hardware depreciation, and financing.
4. Today’s leading builders are financially stronger than many railroad and fiber promoters were
The railroad boom was heavily financed through bonds. When railroad returns declined and European investors withdrew, defaults spread to banks and securities firms and helped produce the Panic of 1873.
The 1990s fiber boom included numerous new or rapidly expanding companies—Global Crossing, Level 3, Qwest and others—whose survival depended on continued access to capital and rapid growth in wholesale bandwidth demand. When demand disappointed, many failed or were reorganized.
The principal AI builders—Amazon, Microsoft, Alphabet and Meta—began with enormous revenues, cash flows, and valuable existing businesses. That makes a wave of immediate hyperscaler bankruptcies considerably less likely than nineteenth-century railroad failures or the failures of many fiber entrants.
The distinction is narrowing, however. Amazon, Alphabet, Meta and Oracle had issued about $194 billion of bonds during 2026 through July 7, 79 percent more than they issued in all of 2025. Goldman Sachs estimated that debt would finance roughly one-third of 2026 hyperscaler capital spending.
Companies are also shifting projects into leases and joint ventures. A $14 billion Meta–BlackRock data-center project in El Paso, for example, will be 80 percent owned by BlackRock-managed funds and partly financed with $12.5 billion of debt, while Meta leases the capacity.
My inference is that a downturn would be more likely to produce project cancellations, asset write-downs, lower returns, and losses for developers, lenders, lessors, utilities, and equipment suppliers than the wholesale disappearance of the leading technology companies. But increasing reliance on debt and outside capital means the financial consequences would no longer be confined to their own shareholders.
5. The return on AI investment is harder to observe
A railroad could measure freight and passenger revenue against the cost of operating a route. A wholesale fiber company could measure the bandwidth it leased and the price per unit of capacity.
Much AI infrastructure is used internally. AI may improve advertising, search, recommendation systems, programming, logistics, customer service, or the defensive position of an existing business. The return may not appear as separately identified “AI revenue.”
That gives the builders greater flexibility, but it also makes investment discipline more difficult. A company may continue spending because it fears falling behind, even when it cannot demonstrate that the next data center will earn an adequate direct return. Meta, for example, derives most of its revenue from advertising while making enormous infrastructure investments intended both to support that business and to develop new AI products.
6. The principal bottleneck is electricity rather than merely construction capital
Railroads competed for land and rights-of-way. Fiber companies needed trenches, conduits, poles, and access to local networks.
AI data centers compete above all for dependable electricity and rapid grid connections. A data center can be completed in two or three years, while new transmission, generation, and other energy infrastructure can require much longer. In some regions, expected data-center loads already exceed available capacity, causing projects to be delayed or paired with on-site natural-gas generation.
This creates a distinctive public-policy problem. Utilities may build generation and transmission for projected data-center demand, only to be left with underused assets if projects are cancelled or use less power than anticipated. Depending on rate design, households and other businesses could bear some of those costs.
7. The form of government assistance is different
Four of the five transcontinental railroads received federal land grants, giving them both rights-of-way and land they could sell to finance construction.
AI data centers have not received an equivalent nationwide transfer of federal land. Assistance has instead generally taken the form of state and local tax concessions, discounted or specially designed electricity tariffs, accelerated permitting, semiconductor subsidies, and publicly regulated grid investment. That creates similar questions about who receives the gains and who bears the costs, but through less visible mechanisms
Which historical analogy is the more useful one?
For the next several years, the fiber analogy is probably more informative. Both involve rapidly improving technology, uncertain demand for applications that do not yet fully exist, extrapolation of very high early growth rates, and the possibility that efficiency improvements will create more usable capacity than planners expected.
For the longer-run physical and economic consequences, the railroad analogy may be more informative. AI is driving investment not only in servers but also in power plants, transmission lines, substations, gas pipelines, batteries, cooling technology, chip factories, and fiber connections. Those investments may alter the economic geography of the United States and remain useful long after the first generation of AI chips has been replaced.
My tentative expectation is therefore a fiber-style shakeout within a railroad-scale infrastructure transformation:
Some announced data centers will never be completed.
Older accelerators will depreciate faster than expected.
Computing prices and profit margins may fall.
The strongest cloud and technology companies will absorb or outlast weaker developers.
Much of the underlying electrical, building, and communications infrastructure will survive and support applications that are not yet apparent.
That is an inference rather than a confident forecast. AI demand might grow quickly enough to absorb most of the planned capacity, especially as lower costs generate more intensive applications. But the historical comparisons show why “AI will transform the economy” and “too much AI capacity is currently being built” are entirely compatible propositions.
Bottom line
The three buildouts share a common economic pattern:
A genuinely revolutionary technology creates a race to build costly infrastructure ahead of uncertain demand. Competition and optimism produce duplication and sometimes overinvestment. Investors may suffer when expectations are revised, yet the surplus infrastructure can lower costs and accelerate the technology’s eventual diffusion.
The largest difference is that railroad rights-of-way and buried fiber are unusually durable, while the most expensive part of an AI data center—the computing equipment—is rapidly depreciating and must be continuously powered. The crucial question is therefore not simply whether demand for AI will grow. It almost certainly will. The question is whether monetizable demand will grow quickly enough to cover the extraordinary depreciation, electricity, and financing costs of the capacity now being installed.
Supports:Microeconomics and Economics, Chapter 14, Section 14.2.
Image created by ChatGPT
An article in the Wall Street Journal discussed why the price of chicken in supermarkets has been falling. The article notes that, “Bigger chicken breeds and flocks not being decimated by disease over the summer have led to a glut in an industry that slaughters more than nine billion birds a year.” The article quotes an industry analyst who is critical of the decisions U.S. poultry famers. According to the analyst, “The industry shot themselves in the foot this year. All you had to do was just be disciplined around production.”
According to the U.S. Department of Agriculture, more than 150,000 farms in the United States sell at least some poultry and eggs, and more than 70,000 farms specialize in selling poultry and eggs.
a. What does the analyst mean by arguing that poultry famers should have been more “disciplined”? What does the analyst expect the result would have been of farmers having been more disciplined?
b. Given the information provided, why might poultry farmers have failed to be more disciplined?
Solving the Problem Step 1: Review the chapter material. This problem is about the difficulty firms have in implicitly colluding if there are many firms in an industry, so you may want to review Chapter 14, Section 14.2, “Game Theory and Oligopoly.”
Step 2: Answer part a. by explaining what that analyst meant by poultry farmers having failed to have been “disciplined” and what he expected the result of farmers being more disciplined would have been. Given the context that U.S. poultry farmers had produced an unusually large number of chickens, the analyst is suggesting that if farmers had been more disciplined, they would have produced fewer chickens. Producing fewer chickens would have reduced the supply of chickens to the market and avoided the decline in chicken prices.
Step 3: Answer part b. by explaining why poultry farmers failed to be more disciplined. The information provided indicates that there are a large number of poultry farmers in the United States. As a result, the quantity of chickens produced by any one farmer is small relative to the total quantity of chickens produced in the market. Therefore, poultry farmers are price takers and no one poultry farmer is able to significantly affect the market price of chicken. (In Chapter 12, Section 12.1, we discuss why firms in a competitive market are price takers.) The only way for poultry farmers to have maintained chicken prices would have been to collude, either explicitly or implicitly, to produce fewer chickens. Explicit collusion is a violation of the antitrust laws and is unlikely to have been effective in any case because individual poultry farmers have a strong incentive to cheat on any agreement to restrict output. The same is true of an attempt by farmers to implicitly collude to restrict supply.
Supports:Macroeconomics, Chapter 14, Section 14.1, Economics, Chapter 24, Section 24.1, and Money, Banking, and the Financial System, Chapter 2, Section 2.1.
Image created by ChatGPT
A rare book dealer who often posts to YouTube made the following observation in one of his videos:
“… thousands of years ago, they had the barter system where you could literally exchange wheat for barley and barley for wheat directly. And the idea behind that was to … have a quick solution for [a] transaction, but over time they invented a monetary unit—coinage and money—and they thought that that would inject some efficiency into economic transactions. And in some ways it’s done the complete opposite. There’s a lot of inefficiency because now unfortunately I cannot go right into Bloomingdale’s and take a nice black suit off the shelf and exchange it for a Geneva Bible. I actually have to sell the Bible first … then go buy the suit. So that gives me a lot of extra work, so I’d rather go back to bartering ….”
The dealer may not have been entirely serious, but assuming that he was, is he correct that transacting using barter is more efficient than transacting using money? In your answer, be sure to define “efficient” in this context.
Solving the Problem Step 1: Review the chapter material. This problem is about the efficiency of using money to purchase goods rather than engaging in barter, so you may want to review Macroeconomics, Chapter 15, Section 15.1, “What Is Money and Why Do We Need It?”
Step 2: Answer the problem by explaining why using money is more efficient than engaging in barter. The book dealer is correct that thousands of years ago, most societies used barter rather than money. Societies transitioned from barter to money because of the inefficiencies of barter. A key inefficiency of barter is the need for a double coincidence of wants. For a barter transaction to take place, each person must want what the other person has. It’s not enough for the book dealer to want a black suit from the Bloomingdale’s department store; Bloomingdale’s must be willing to trade the suit for a copy of the Geneva Bible—which is unlikely.
To use a copy of the Geneva Bible to obtain a suit using barter, the book dealer might have to make—possibly many—additional trades until he obtains some good that Bloomingdale’s would accept in exchange for the suit. In practice, it might be difficult to find such a good and doing so would likely involve substantial search costs.
We can conclude that money has replaced barter in most transaction because it is more efficient in the sense that it allows transactions to be completed at a lower cost.
The following was the first sentence of an article yesterday on axios.com discussing the market for beef: “Beef sales are plunging, but processors continue to raise prices as a yearslong cattle shortage strains the industry.”
The sentence seems to be describing a paradox: Why would meat processors, such as Tyson, JBS, and Cargill, raise beef prices if their sales are falling? The key to resolving the apparent paradox is the reference to a “cattle shortage.” The number of cattle raised in the United States has been declining for several reasons, including severe drought in cattle-raising states—which has reduced the pasture that cattle forage on—and a reduction in beef imports from Mexico as the United States Department of Agriculture (USDA) tries to limit the spread of screwworm.
In other words, using the model of demand and supply we develop in Chapter 3 of Microeconomics, the supply curve for beef in the United States has shifted to the left. The result is shown in the following figure:
When the supply curve shifts to the left from S1 to S2, the price of beef rises from P1 to P2 and the equilibrium quantity of beef falls from Q1 to Q2. In other words, when a market experiences a decline in supply, we would expect to observe both higher prices and falling sales. So, the situation described in the first sentence of the article is not a paradox, but instead reflects the normal working of demand and supply in a market. You can explain a lot just by knowing that demand curves slope downward!
The article also observes with respect to Tyson Foods that: “In its most recent quarter, ended June 27, beef volumes declined by 15.9% from a year ago, while prices Tyson charged grocery stores, restaurants and other customers rose 12.1%.” The USDA estimates that the retail price elasticity of demand for beef is about –1. If we assume that no other factors affecting the demand for Tyson’s beef changed during this three-month period, then the price elasticity of demand for Tyson’s beef is –15.9%/12.1% = –1.3. (Note that the USDA elasticity estimates are for beef sold in supermarkets and other retail venues. So the estimates may not directly apply to sales to restaurants and “other customers.”)
We would expect that the price elasticity of demand for Tyson’s beef would be larger (in absolute value) than the price elasticity of demand for beef as a good. As we discuss in Chapter 6 of Microeconomics, if the price of one brand of a good increases, consumers can switch to another brand. In this case, if the price of Tyson’s beef increases, some consumers will switch to Cargill’s or some other firm’s beef. But if the price of beef as a good increases, consumers would have to eat a different protein to avoid the price increase.
This morning (July 2)—one day early because tomorrow is a federal holiday—the Bureau of Labor Statistics (BLS) released its “Employment Situation” report (often called the “jobs report”) for June. The report showed a smaller than expected increase in employment.
The jobs report has two estimates of the change in employment during the month: one estimate from the establishment survey, often referred to as the payroll survey, and one from the household survey. As we discuss in Macroeconomics, Chapter 9, Section 9.1 (Economics, Chapter 19, Section 19.1), many economists and Federal Reserve policymakers believe that employment data from the establishment survey provide a more accurate indicator of the state of the labor market than do the household survey’s employment and unemployment data. (The groups included in the employment estimates from the two surveys are somewhat different, as we discuss in this post.)
According to the establishment survey, there was a net increase of 57,000 nonfarm jobs during June. Economists surveyed by the Wall Street Journal had forecast an increase of 115,000 jobs. Economists surveyed by FactSet had a lower forecast of a net increase of 100,000 jobs. The BLS revised downward its previous estimates of employment in April and May by a combined 74,000 jobs. (The BLS notes that: “Monthly revisions result from additional reports received from businesses and government agencies since the last published estimates and from the recalculation of seasonal factors.”)
The following figure from the jobs report shows the net change in nonfarm payroll employment for each month in the last two years. The figure shows that the relatively strong 137,000 average net increase in jobs over the past four months represents a break from the unusual pattern in that began in the middle of 2025 in which months of declining employment and months of increasing employment had been alternating.
These employment gains conflict with a popular view among economists that slowing labor force growth has driven the break-even rate of employment growth—the rate required to keep the unemployment rate constant—down to nearly zero
Despite the relatively small increase in employment in June, the unemployment rate, which is calculated from data in the household survey, declined to 4.2 percent from 4.3 percent in May at 4.3. The decline in the unemployment rate was due to a decline in the estimated size of the labor force, an estimate that fluctuates significantly from month to month. Despite that fact, as the following figure shows, the unemployment rate has been remarkably stable over the past year and a half, staying between 4.0 percent and 4.4 percent in each month since May 2024. The Federal Open Market Committee’s current estimate of the natural rate of unemployment—the normal rate of unemployment over the long run—is 4.2 percent. So, currently the unemployment rate is equal to that estimate of the natural rate. (We discuss the natural rate of unemployment in Macroeconomics, Chapter 9 and Economics, Chapter 19.)
As the following figure shows, the monthly net change in jobs from the household survey moves much more erratically than does the net change in jobs from the establishment survey. As measured by the household survey, there was a net decrease of 507,000 jobs in June, as compared to the net increase in employment shown in the establishment survey. In addition, the household survey shows a significant net decline in jobs during the past six months, in contrast to the significant net increase in jobs shown in the establishment survey. (Note that because of last year’s shutdown of the federal government, there are no data for October or November.)
The household survey has another important labor market indicator: the employment-population ratio forprime age workers—those workers aged 25 to 54. In June. the ratio declined sharply to 80.2 percent from 80.8 percent in May, the lowest value since December 2022. The decline in the prime-age population ratio is difficult to reconcile with the net increase in employment shown in the payroll survey. The state of the labor market in June seemed significantly weaker in household survey data than in establishment survey data.
There have been media reports of firms, including Salesforce, Cloudflare, Coinbase, Cisco Systems, and Meta Platforms, laying off workers in information systems. The following figure shows net employment changes in the BLS employment category of “computing infrastructure providers, data processing, web hosting, and related services.” Employment in this sector has been declining during most months since the beginning of 2023. June was no exception with a net decrease of 3,300 jobs.
The establishment survey also includes data on average hourly earnings (AHE). As we noted in this post, many economists and policymakers believe the employment cost index (ECI) is a better measure of wage pressures in the economy than is the AHE. The AHE does have the important advantage of being available monthly, whereas the ECI is only available quarterly. The following figure shows the percentage change in the AHE from the same month in the previous year. The AHE increased 3.5 percent in June, up slightly from 3.4 percent in May.
What effect is this jobs report likely to have on the decisions of the Federal Reserve’s policymaking Federal Open Market Committee (FOMC) at its next meeting on July 28–19? The slowdown in employment growth reduces the chance that the FOMC will increase its target range for the federal funds rate. The probability that investors in the federal funds futures market assign to the FOMC increasing its target range at that meeting fell from 28.9 percent yesterday to 17.6 percent this morning. Investors still assign a 54.0 percent probability to the FOMC raising its target range at its September meeting, but that was down from 64.1 percent yesterday.
Supports: Microeconomics and Economics, Chapter 6, Section 6.3, and Essentialsof Economics, Chapter 7, Section 7.7.
Image created by ChatGPT
An article in the Wall Street Journal on June 25, noted that after Apple increased the prices of iPads and MacBooks, the price of its stock declined by 6.1 percent. That decline meant that the total value of Apple’s stock—its market cap—fell by $215 billion dollars that day. Investors were expecting that Apple would likely also increase the prices of iPhones. As we discuss in Microeconomics, Chapter 8 (Macroeconomics and Essentials of Economics, Chapter 6), the price of a firm’s stock reflects investors forecasts of the future profitability of the firm. Why would Apple increasing the prices of its products cause investors to believe that Apple’s profit would decline? Shouldn’t Apple become more profitable after increasing its prices?
Solving the Problem Step 1: Review the chapter material. This problem is about the effect on a firm’s profit of increasing the price of its product, so you may want to review Chapter 6, Section 6.3, “The Relationship between Price Elasticity of Demand and Total Revenue.”
Step 2: Answer the question by explaining under what circumstances a firm may reduce its profit by raising prices. It might make sense to think that any time a firm raises its price, it will increase its profit. But recall that because demand curves slope downward, an increase in price always results in a decrease in the quantity of the good sold. If the firm’s demand curve is elastic at the current price level, raising the price will decrease the firm’s revenue because the quantity sold will fall by proportionally more than the price increases. In this case, investors appear to have assumed that the revenue Apple would lose as a result of raising prices would be greater than the additional revenue it would earn on the quantities it would sell at the higher prices. Revenue isn’t the same as profit because Apple’s total cost will decrease as it sells a smaller quantity. Because the price of Apple’s stock declined substantially on the day the firm announced the price increases, investors must be expecting that the net effect of the price increases would be to reduce Apple’s profit.
