Edward Conard

Top Ten New York Times Bestselling Author

  • “Unintended Consequences offers deep and well-argued analyses on almost every issue.” - The New York Times
  • “…reminds us that inequality sends a signal of what society lacks most, in America’s case, entrepreneurship and risk taking.” - Lawrence Lindsey, CEO, The Lindsey Group, former Director of the National Economic Council
  • “…a fresh argument for the productive value of inequality.” - David Autor, Professor of Economics, Massachusetts Institute of Technology
  • “…challenges misconceptions that distort our economic debates.” - Arthur Brooks, President of the American Enterprise Institute
  • “…a fresh argument for the productive value of inequality.” - David Autor, Professor of Economics, Massachusetts Institute of Technology
  • “…serious thinking for serious thinkers. …a thought-provoking blueprint for growing middle- and working-class incomes.” - Mitt Romney, former Governor of Massachusetts
  • “…challenges misconceptions that distort our economic debates.” - Arthur Brooks, President of the American Enterprise Institute
  • “…a very valuable contribution.” - Larry Summers, former Secretary of the Treasury and director of the National Economic Council, president emeritus, Harvard University
  • “…a must-read for serious students of economic policy.” - Glenn Hubbard, Dean, Columbia Business School, and former Chairman of the Council of Economic Advisers
  • “…reminds us that inequality sends a signal of what society lacks most, in America’s case, entrepreneurship and risk taking.” - Lawrence Lindsey, CEO, The Lindsey Group, former Director of the National Economic Council
  • “…a comprehensive explanation of the modern economy.” - Julian Robertson, Founder, Tiger Management
  • “…a comprehensive explanation of the modern economy.” - Julian Robertson, Founder, Tiger Management
Upside of Inequality Oxford Unintended Consequences

Macro Roundup


Monday, February 24, 2025

AI Summary. SWE-Lancer is a benchmark of over 1,400 real freelance software engineering tasks worth $1m in actual payouts, spanning $50 bug fixes to $32,000 feature builds plus managerial choices between technical proposals. Evaluations find leading AI models still cannot solve the majority of the tasks.

Samuel Miserendino, Michele Wang, Tejal Patwardhan and Johannes Heidecke arXiv
Date Posted:

AI programs are making progress in solving freelance software engineering tasks, but they are still less capable than experienced engineers. The top-performing model successfully completed tasks worth $208k on Upwork, out of 1,400 tasks worth $1mm.

Can leading AI models complete most real-world freelance software engineering tasks?

Core argument: SWE-Lancer grades independent engineering tasks with end-to-end tests triple-verified by experienced software engineers, and scores managerial tasks by comparing model choices among technical proposals against the decisions of the originally hired engineering managers.

We introduce SWE-Lancer, a benchmark of over 1,400 freelance software engineering tasks from Upwork, valued at $1 million total in real-world payouts. SWE-Lancer encompasses both independent engineering tasks — ranging from $50 bug fixes to $32,000 feature implementations — and managerial tasks, where models choose between technical implementation proposals. Independent tasks are graded with end-to-end tests triple-verified by experienced software engineers, while managerial decisions are assessed against the choices of the original hired engineering managers. We evaluate model performance and find that frontier models are still unable to solve the majority of tasks. To facilitate future research, we open-source a unified Docker image and a public evaluation split. By mapping model performance to monetary value, we hope SWE-Lancer enables greater research into the economic impact of AI model development.

Takeaways by Macro Roundup® AI

  1. SWE-Lancer grades independent engineering tasks with end-to-end tests triple-verified by experienced software engineers, and scores managerial tasks by comparing model choices among technical proposals against the decisions of the originally hired engineering managers.
  2. The authors open-source a unified Docker image and a public evaluation split, and map model performance to monetary value to support research into the economic impact of AI model development.

Related Articles:

  • Society’s Technical Debt and Software’s Gutenberg Moment — Every wave of technological innovation has been unleashed by something costly becoming cheap enough to waste, and large language models are collapsing the cost and complexity of producing software. The authors argue that decades of underproduced software have left society with immense technical debt that is poised to contract, releasing a wave of innovation.
  • The Short-Term Effects of Generative Artificial Intelligence on Employment: Evidence from an Online Labor Market — Freelancers in occupations most exposed to generative AI saw employment and earnings decline, with the authors estimating a 2% drop in jobs and a 5.2% drop in monthly earnings on an online labor platform. Offering high-quality service did not mitigate the effect, and top-rated freelancers appear disproportionately hurt.
  • The Rise—and Fall—Of The Software Developer — An index of employment at 6,500 companies tracking more than 75,000 software developers and engineers found developer employment grew through late 2019 and has been falling since 2020. Over the same period, developers' median base pay rose 24% versus 30% for all U.S. workers, though developers still earn more than 2x the U.S. median salary.

AI Summary. The authors estimate that labor demand falls with the average exposure of workers' tasks to artificial intelligence but rises with the dispersion of task exposures, as workers shift effort toward tasks AI does not displace. Firm productivity gains and task reallocation nearly offset direct substitution, leaving the net effect near zero across most of the pay distribution.

Menaka Hampole, Dimitris Papanikolaou, Lawrence Schmidt and Bryan Seegmiller National Bureau of Economic Research
Date Posted:

.@menaka_hampole et al. estimate that AI has little net effect on overall labor demand. Though “highly-exposed occupations experience relatively lower demand…the resulting increase in firm productivity increases employment across all occupations.”

How does AI exposure affect labor demand across the pay distribution?

Core argument: Labor demand falls with the average exposure of workers' tasks to AI but rises with the dispersion of task exposures, as workers shift effort toward tasks that AI does not displace.

The impact on wage earnings and employment can be summarized by two statistics: [1] Labor demand decreases in the average exposure of workers' tasks to AI technologies; [2] holding the average exposure constant, labor demand increases in the dispersion of task exposures to AI, as workers shift effort to tasks that are not displaced by AI. The direct effect of labor substitution - the red line in the top panel of Figure 4 [is that] the negative effect of mean AI exposure increases on average as pay increases, echoing the fact that the AI exposure is higher for higher-paid occupations. However, this mean exposure measure abstracts from the benefits of labor reallocation and the impact of firm productivity on firm labor demand. Examining the remaining lines in the Figure, we note these two effects serve to mediate the direct effect of labor substitution. Highly-paid positions tend to be at firms which use AI intensively (in yellow), and are also exposed to larger dispersion in AI exposure (in green). These two forces almost offset the impact of direct task-level AI substitution, so that the total net effect in blue is close to zero [except at the top of] the pay distribution.

Takeaways by Macro Roundup® AI

  1. Labor demand falls with the average exposure of workers' tasks to AI but rises with the dispersion of task exposures, as workers shift effort toward tasks that AI does not displace.
  2. The direct labor-substitution effect of AI grows more negative on average as pay increases, echoing the higher AI exposure of higher-paid occupations.
  3. Higher-paid positions tend to sit at firms that use AI intensively and face larger dispersion in AI exposure, forces that almost offset direct task-level substitution, leaving the net effect close to zero except at the top of the pay distribution.

Related Articles:

  • The Rapid Adoption of Generative AI — A nationally representative U.S. survey found 39% of people aged 18-64 used generative AI, with more than 24% of workers using it in the prior week and nearly one in nine using it every workday. The authors estimate 0.5-3.5% of U.S. work hours are currently assisted by generative AI.
  • Use of Artificial Intelligence and Productivity: Evidence from Firm and Worker Surveys — A survey of Japanese firms and workers finds that firms using artificial intelligence have higher productivity, mean wages, and expected medium-term growth, and that more-educated firms and workers are more likely to use it. Workers using artificial intelligence report their own productivity rose by about 20% on average.
  • Technological Disruption in the Labor Market — The authors measure US labor market structure since 1880 and find 1990–2017 less disruptive than any earlier period, because past changes from general-purpose technologies like steam and electricity were so profound. They identify four recent signs of accelerating change, including a 50%+ rise in science and technology employment share since 2010.

AI Summary. Federal workforce layoffs and contract cuts are projected by consensus to eliminate 300,000 jobs, but with studies showing two contractors for every federal employee, total job losses could approach 1mm. Economic policy uncertainty is spiking while credit spreads have not widened and incoming data remain strong.

Torsten Sløk Apollo
Date Posted:

Noting a disconnect between credit spreads and an index of policy uncertainty, Apollo Chief Economist Torsten Sløk warns that downside risks to capex and hiring decisions are growing.

Are federal job and contract cuts a growing risk to the economy?

Core argument: Consensus expects 300,000 DOGE-related job cuts, small against U.S. employment of 160mm, 7mm unemployed, and roughly 5mm people changing jobs each month.

We are starting to worry about the downside risks to the economy and markets from the impact of DOGE layoffs and contract cuts, weighing on capex spending decisions and hiring decisions. The consensus expects total DOGE-related job cuts to be 300,000. Total employment in the United States is 160mm, with 7mm unemployed. About 5mm people change jobs every month. In that context, 300,000 federal jobs lost is not much. Studies show that for every federal employee, there are two contractors. As a result, layoffs could potentially be closer to 1 million. Any increase in layoffs will push jobless claims higher over the coming weeks, and such a rise in the unemployment rate is likely to have consequences for rates, equities, and credit. Economic policy uncertainty is spiking higher, but credit spreads are not widening. The question is if persistently elevated policy uncertainty will begin to have a negative impact on capex spending and hiring decisions. The incoming data remain strong. But the near-term downside risks to the economy and markets are growing.

Takeaways by Macro Roundup® AI

  1. Consensus expects 300,000 DOGE-related job cuts, small against U.S. employment of 160mm, 7mm unemployed, and roughly 5mm people changing jobs each month.
  2. Studies showing two contractors for every federal employee suggest DOGE-related layoffs could be closer to 1mm, which would push jobless claims higher over the coming weeks.
  3. Economic policy uncertainty is spiking while credit spreads are not widening and incoming data remain strong, though a rising unemployment rate would likely have consequences for rates, equities, and credit.

Related Articles:

  • When History Rhymes — Inflation, inflation expectations, and the gap between nominal and inflation-indexed bond yields are all moving higher. The author argues that cutting interest rates too early raises the likelihood of repeating the 1970s, leaving the Federal Reserve no option but to hold rates higher for longer.
  • Are We Still on the Path to Equilibrium? — If the Federal Reserve delivers the normalization toward its projected long-term neutral policy rate, market-cap-weighted US asset yields would offer a near-normal risk premium over cash. If instead demand stabilizes at a higher level with tight capacity and fading disinflationary forces, policy could not ease as much as markets discount, so asset yields would likely rise and prices fall.
  • Business Spending Is Lining Up to Reinforce Growth — Around 70% of sectors are posting positive year-over-year profit growth, with big tech showing the largest acceleration in demand and profits but strength broadening across the economy. Companies are responding with rising capital spending growth and more optimistic investment plans, as firms typically add capacity with a slight lag after real demand picks up.

AI Summary. Microsoft has voided data center leases totaling a couple of hundred megawatts of capacity and stopped converting preliminary agreements into formal leases, according to brokerage TD Cowen's checks with supply chain providers. TD Cowen also says Microsoft is redirecting some planned international spending to the US, suggesting a material slowdown in international leasing.

Ryan Vlastelica and Newley Purnell Bloomberg
Date Posted:

Is Microsoft growing more cautious about demand for AI? A TD Cowen report found that the company has started canceling leases for a “substantial” amount of data center capacity.

Is a major tech company pulling back on AI data center leases?

Core argument: Microsoft has voided AI data center leases totaling "a couple of hundred megawatts" of capacity, according to TD Cowen, which cited channel checks with supply chain providers.

Microsoft OpenAI’s biggest backer has voided leases totaling “a couple of hundred megawatts” of capacity, TD Cowen reported, citing channel checks or inquiries with supply chain providers. Microsoft has also stopped converting agreements that usually lead to formal leases, TD Cowen said. That was a tactic rivals such as Meta Platforms Inc. employed previously, when it decided to cut back on capital spending, the brokerage wrote. Microsoft is redirecting a portion of its planned international spending to the US too, TD Cowen said, which “suggests to us a material slowdown in international leasing.” A potential pullback by Microsoft on spending and datacenter construction raises questions about whether the company — one of the frontrunners among Big Tech in AI — is growing cautious about the outlook for demand. The company has said it expects to spend $80 billion this fiscal year on AI data centers, and, on a late January earnings call, Chief Executive Officer Satya Nadella said Microsoft has to sustain spending to meet “exponentially more demand.”

Takeaways by Macro Roundup® AI

  1. Microsoft has voided AI data center leases totaling "a couple of hundred megawatts" of capacity, according to TD Cowen, which cited channel checks with supply chain providers.
  2. TD Cowen said Microsoft has stopped converting agreements that usually lead to formal leases, a tactic Meta Platforms used previously when it decided to cut back on capital spending.
  3. Microsoft is redirecting a portion of planned international spending to the US, which TD Cowen said suggests a material slowdown in international leasing, even as Microsoft has said it expects to spend $80 billion this fiscal year on AI data centers.

Related Articles:

  • Gen AI: Too Much Spend, Too Little Benefit? — The author argues generative AI is unlikely to drive revenue expansion because it helps employees find information faster rather than find better information. If AI's benefits remain limited to efficiency gains, competitors can adopt the same tools, so cost savings get arbitraged away rather than raising prices or margins.
  • A Severe Case of COVIDIA: Prognosis For An AI-Driven US Equity Market — Computing cycles move from infrastructure to platforms to applications, and NVIDIA data center revenues as a share of market-wide capital spending are projected to reach levels last seen at the 1969 mainframe peak and the dot-com boom. Large cloud providers spend more than twice as much training foundational models as running them for customers.
  • The AI Transition One Year Later: On Track, but Macro Impact Still Several Years Off — Investment in AI hardware is projected to rise substantially as a precursor to integrating AI into business processes, with the investment cycle estimated to peak at 2–2½% of GDP and early signs visible among AI-enabling companies. The authors estimate generative AI could raise developed-market labor productivity growth by 1.5pp annually over a 10-year adoption period, roughly 15% cumulatively.

AI Summary. Moody's Analytics estimates the top 10% of earners account for 49.7% of all U.S. consumer spending, a record in data going back to 1989, up from about 36% three decades ago, with their spending alone equal to almost one-third of gross domestic product.

Rachel Louise Ensign Wall Street Journal
Date Posted:

The top 10% of US earners now account for nearly half of all personal spending, according to Moody’s @Markzandi, up from 36% three decades ago. “The finances of the well-to-do have never been better… and the economy never more dependent on that group.”

How much of US consumer spending comes from the top 10%?

Core argument: The top 10% of earners account for 49.7% of all U.S. consumer spending, a record in data going back to 1989 and up from about 36% three decades ago, according to a Moody's Analytics analysis of Federal Reserve data.

Mark Zandi estimated that spending by the top 10% alone accounted for almost one-third of gross domestic product. Those consumers now account for 49.7% of all spending, a record in data going back to 1989, according to an analysis by Moody’s Analytics. Three decades ago, they accounted for about 36%. Between September 2023 and September 2024, the high earners increased their spending by 12%. Spending by working-class and middle-class households, meanwhile, dropped over the same period. The bottom 80% of earners spent 25% more than they did four years earlier, barely outpacing price increases of 21% over that period. The top 10% spent 58% more. "The finances of the well-to-do have never been better, their spending never stronger and the economy never more dependent on that group,” said Zandi, who oversaw the analysis, which was based on data from the Federal Reserve. The analysis runs through the third quarter of 2024 because that is the most recent data available.

Takeaways by Macro Roundup® AI

  1. The top 10% of earners account for 49.7% of all U.S. consumer spending, a record in data going back to 1989 and up from about 36% three decades ago, according to a Moody's Analytics analysis of Federal Reserve data.
  2. Mark Zandi estimated that spending by the top 10% alone accounted for almost one-third of U.S. gross domestic product, and said the economy has never been more dependent on that group.
  3. Between September 2023 and September 2024, high earners raised their spending 12% while working-class and middle-class household spending fell; over four years the bottom 80% spent 25% more against 21% price increases, versus 58% for the top 10%.

Related Articles:

  • The American Dream Is Not a Coin Flip, and Wages Have Not Stagnated — Using an alternative cost-of-living adjustment, the typical 40-year-old's pre-tax family income rose 52% between 1979 and 2019, or 73% adjusting for family size, versus 15% under the methods behind widely cited mobility estimates. Absolute mobility is down 20 percentage points since 1979 but stood at 72% in 2023.
  • The Distribution of Household Income in 2021 — Income after transfers and taxes grew most for households in the highest quintile, a rate that would have been nearly the same without temporary pandemic policies such as recovery rebate credits, expanded unemployment compensation, and the expanded child tax credit. Those policies raised average income in the lowest quintile to $48,700 from $41,200 without them.
  • Changing Opportunity: Sociological Mechanisms Underlying Growing Class Gaps and Shrinking Race Gaps in Economic Mobility — Across the 1978 to 1992 birth cohorts, adult household incomes fell for white children raised in low-income families and rose for those raised in high-income families, increasing the intergenerational persistence of income ranks among white children by 28%. Over the same cohorts, adult incomes rose for Black children at all parental income levels.

AI Summary. Among the world's 2,000 largest companies, U.S. firms generated 38% of global profits and firms based in allied countries 35%, while Chinese firms including Hong Kong generated 16%. The U.S. leads in 20 of 27 industries and China in three, with profits used as a measure of economic chokepoints.

Stephen Brooks and Ben Vagle Foreign Affairs
Date Posted:

China is far more vulnerable than the US to bilateral trade disruptions, argue Stephen Brooks & @ben_vagle. In 2022, US firms demonstrated significantly greater economic power, with 38% of global profits vs. 16% for Chinese firms.

How do global corporate profits compare between the US and China?

Core argument: U.S. firms generated 38% of the profits of the 2022 Forbes 2000, with allied-country firms at 35% and Chinese firms, including those in Hong Kong, at 16%.

Look at the profits generated by the 2022 Forbes 2000—the world’s 2,000 largest companies. Profits are the preferred measure of economic power because if a firm in a sector is generating them, it likely means there are barriers preventing competitors from entering the market and cutting into that company’s margins. They thus best capture the chokepoints of the world economy. And U.S. firms generated 38% of global profits, while firms headquartered in allied countries generated 35%. Chinese firms, including those in Hong Kong, generated just 16%. A closer look at the 27 industries in the Forbes 2000 makes the U.S. lead over China even clearer. China leads in three of these industries. The United States, meanwhile, leads in 20, of them, almost always by double digits. In three of the seven industries in which the United States is not the leader, an American ally is. Together, the United States and its allies and partners make up all the top five countries in terms of profit share in five industries: aerospace and defense, drugs and biotechnology, media, semiconductors, and utilities.

Takeaways by Macro Roundup® AI

  1. U.S. firms generated 38% of the profits of the 2022 Forbes 2000, with allied-country firms at 35% and Chinese firms, including those in Hong Kong, at 16%.
  2. The United States leads in 20 of the 27 industries covered by the Forbes 2000, almost always by double digits, while China leads in three, and an American ally leads three of the remaining seven.
  3. The United States and its allies and partners occupy all top five country slots by profit share in aerospace and defense, drugs and biotechnology, media, semiconductors, and utilities.

Related Articles:

  • US Efforts to Contain Xi’s Push for Tech Supremacy Are Faltering — Research by Bloomberg Economics and Bloomberg Intelligence finds that of 13 key technologies tracked, China has reached global leadership in five and is catching up in seven others, with clear leads in electric vehicles, automotive software and lithium battery technology. China still struggles to develop manufacturing processes for advanced semiconductors, the main focus of US export controls.
  • Crises at Boeing and Intel Are a National Emergency — Intel and Boeing, once benchmarks for high-quality, demanding manufacturing, declined because of their own mistakes rather than cheap foreign competition. Because leading U.S. tech firms' software and devices depend on advanced semiconductors fabricated under contract, especially by Taiwan Semiconductor Manufacturing Co., Chinese subjugation of Taiwan could leave the U.S. tech sector at Beijing's mercy.
  • 2023 Letter — Rising numbers of Chinese nationals are being apprehended at the US-Mexico border, with US officers encountering around 1,000 in one month early in the year and nearly 5,000 by November. Most Chinese emigrants leave through legal means, going primarily to Singapore, Japan, and Thailand.

AI Summary. Physicists outside the work have responded cautiously to Microsoft's topological quantum computing claim, noting that the search for topological qubits and the "Majorana" particles underpinning them has a history of bold announcements later withdrawn. Researchers describe the technology as excruciatingly difficult to actually build.

Nidhi Subbaraman Wall Street Journal
Date Posted:

Microsoft’s quantum computing breakthrough in creating an advanced “topological qubit” is likely to prove difficult to turn into practical outcomes, some physicists contend.

Are physicists skeptical of claims of a topological quantum computing breakthrough?

Core argument: Physicists are responding cautiously to Microsoft's quantum announcement, in part because the pursuit of topological qubits and the "Majorana" particles underpinning them has a history of bold pronouncements later pulled back, said Vlad Pribiag, a quantum device physicist at the University of Minnesota.

Scientists are responding to Microsoft’s announcement with caution, in part because the quest for topological qubits and the “Majorana” particles that underpin them has a history of bold pronouncements that later were pulled back, said Vlad Pribiag, a quantum device physicist at the University of Minnesota who wasn’t involved with the work. “This particular technology they’re pursuing, as exciting as it is, is turning out to be excruciatingly difficult to actually build,” said Stephen Bartlett, director of the University of Sydney Nano Institute, who wasn’t involved with the work.

Takeaways by Macro Roundup® AI

  1. Physicists are responding cautiously to Microsoft's quantum announcement, in part because the pursuit of topological qubits and the "Majorana" particles underpinning them has a history of bold pronouncements later pulled back, said Vlad Pribiag, a quantum device physicist at the University of Minnesota.
  2. Stephen Bartlett, director of the University of Sydney Nano Institute, said the technology Microsoft is pursuing, as exciting as it is, is turning out to be excruciatingly difficult to actually build.

Related Articles:

  • Microsoft’s Majorana 1 Chip Carves New Path for Quantum Computing — Microsoft unveiled Majorana 1, a quantum chip built on a Topological Core architecture using a new materials stack of indium arsenide and aluminum, which the company expects will enable quantum computers capable of solving industrial-scale problems within years rather than decades. The topological superconductor produces a qubit that is fast, small, stable, and digitally controlled.
  • Interferometric Single-Shot Parity Measurement In InAs–Al Hybrid Devices — Researchers demonstrated a single-shot interferometric measurement of fermion parity in indium arsenide–aluminium hybrid devices with a gate-defined superconducting nanowire, achieving a parity assignment error probability of 1%. The authors note the results admit both topologically trivial and non-trivial interpretations, and the architecture is compatible with future tests of anyon fusion rules.
  • Meet Willow, Our State-of-the-Art Quantum Chip — Google reports that its Willow quantum chip completed a random circuit sampling benchmark computation in under five minutes that one of today's fastest supercomputers would take an estimated 10^25 years to perform. The benchmark is described as the classically hardest test currently runnable on a quantum computer.

Tuesday, February 25, 2025

AI Summary. A modeled 25% tariff on all automotive, pharmaceutical, and semiconductor imports is estimated to raise the average effective tariff rate by 5.2 percentage points, to its highest level since 1965, and to raise $980bn over a decade without retaliation or $900bn with it. The authors project lower US real output in both the short and long run.

Ernie Tedeschi Yale University
Date Posted:

Trump’s proposed 25% tariffs on autos, pharma, and semiconductors could raise more than $900 billion in revenue over the next 10 years but would cost the economy roughly $30 billion a year in lost output. @ernietedeschi

How much revenue would sector-specific tariffs raise, and at what output cost?

Core argument: The Budget Lab estimates a 25% tariff on all automotive, pharmaceutical, and semiconductor imports would raise the average effective tariff rate by 5.2 percentage points, to its highest level since 1965.

The Budget Lab modeled a 25% tariff on all automotive, pharmaceutical, and semiconductor imports. The illustrative proposal is the equivalent of a rise in the average effective tariff rate of 5.2pp, bringing it to its highest level since 1965. Were it to remain in place, the policy would raise $980 billion over 2026-35 conventionally-scored without retaliation (0.3% of GDP on average over 10 years), and $900 billion with retaliation (0.2% of GDP). The proposal reduces the short- and long-run level of US real GDP. In calendar year 2025, real GDP growth is 0.2pp lower without retaliation and 0.3 percentage point with. In the long-run, US output is still -0.8 to -0.12% lower without retaliation and almost twice as much with. That’s the equivalent of the US economy being permanently smaller by $23-$35 billion annually in 2024 dollars.

Takeaways by Macro Roundup® AI

  1. The Budget Lab estimates a 25% tariff on all automotive, pharmaceutical, and semiconductor imports would raise the average effective tariff rate by 5.2 percentage points, to its highest level since 1965.
  2. The Budget Lab projects the tariffs would raise $980 billion over 2026–35 conventionally scored without retaliation (0.3% of GDP on average over 10 years) and $900 billion with retaliation (0.2% of GDP).
  3. The Budget Lab estimates 2025 real GDP growth would be 0.2 percentage point lower without retaliation and 0.3 percentage point lower with it, leaving the US economy permanently smaller by $23–$35 billion annually in 2024 dollars.

Related Articles:

  • A User’s Guide to Restructuring the Global Trading System — During the 2018–2019 U.S.–China trade war, the effective tariff rate on Chinese imports rose 17.9 percentage points while the renminbi depreciated 13.7% against the dollar, leaving after-tariff import prices in dollars up just 4.1%. The author argues this currency offset of more than three-fourths of the tariff explains why inflation pressure was negligible.
  • Tariffs and Exchange Rates (and Stephen Miran) — The author argues that a universal tariff would prompt enough dollar appreciation to make US exports more expensive abroad, reducing trade flows in both directions while leaving the trade deficit roughly unchanged. Attempting to offset the export drag through deregulation would shift production away from import-competing goods, again leaving a similar trade deficit.
  • Trump Widens Trade Fight to Include Global Taxes and Regulation — The US has directed economic officials to calculate new country-by-country "reciprocal" tariffs based on the total tariff, tax, regulatory, currency and other barriers facing US exports, treating value-added taxes as a trade barrier. If value-added tax rates are added to existing tariff rates, India would face the highest rate at 29%, followed by Brazil and the European Union.

AI Summary. Using micro data on rental properties, the authors document that low-rent housing earns substantially higher total returns than high-rent housing, with annual returns 1.74% higher in Belgium, 3.6% in the Netherlands, and 3.86% in the United States. They argue risk-based explanations cannot account for the full gap, pointing instead to limits to arbitrage in the low-end rental market.

Sven Damen, Matthijs Korevaar, and Stijn Van Nieuwerburgh National Bureau of Economic Research
Date Posted:

Despite significantly higher returns for low-income housing in the US and two European countries, institutional investors shy away from the market, limiting investment and leaving poor tenants facing steeper rents. @damensven @SVNieuwerburgh

Does low-rent housing earn higher returns than high-rent housing?

Core argument: Annual total returns on rental housing are higher for low-rent than high-rent properties by 1.74% in Belgium, 3.6% in the Netherlands, and 3.86% in the United States, according to the authors' micro data.

We document a strong negative relationship between the level of rents and the return to rental housing. An investment in houses in the cheapest 10% of the market earns returns that are substantially higher than an investment in the most expensive 10%. Using high-quality micro data, we show that this relationship holds robustly across housing market settings and time periods. It is present in gross and net rental yields as well as in capital gain yields. Combining net yields with capital gain yields, we arrive at the total return on rental properties. Annual returns are 1.74% higher in Belgium, 3.6% in the Netherlands, and 3.86% in the United States for low-rent than high-rent properties. Investors in low-end properties earn significantly higher returns. We argue that natural risk-based explanations such as systemic risk, regulatory risk, or idiosyncratic cash-flow risk associated with underdiversified portfolios are unlikely to explain the entire return gap. Rather, there appear to be limits to arbitrage that plague the low-tier rental market segment. Most landlords display strong local bias and lack the resources to scale up due to financial constraints. The largest institutional landlords who could arbitrage away the returns are held back by reputation concerns and potential diseconomies of scale. This equilibrium allows for medium-scale geographically-concentrated investors to earn excess returns, at the expense of low-income tenants.

Takeaways by Macro Roundup® AI

  1. Annual total returns on rental housing are higher for low-rent than high-rent properties by 1.74% in Belgium, 3.6% in the Netherlands, and 3.86% in the United States, according to the authors' micro data.
  2. The authors document a strong negative relationship between rent levels and rental housing returns, with houses in the cheapest 10% of the market earning substantially higher returns than the most expensive 10%, in gross and net rental yields and capital gain yields.
  3. The authors argue that risk-based explanations such as systemic, regulatory, or idiosyncratic cash-flow risk are unlikely to explain the entire return gap, pointing instead to limits to arbitrage from landlords' local bias, financial constraints, and institutional landlords' reputation concerns.

Related Articles:

  • The Determinants Of Declining Internal Migration — A 10% increase in wages where people live is associated with 3.5% less migration, while a 10% increase in destination wages is associated with 7.8% more migration; a 10% rise in destination home prices is associated with 2.6% less migration. The authors find wage responses alone would have raised migration rates, but housing-related factors more than offset them.
  • Where is Standard of Living the Highest? Local Prices and the Geography of Consumption — The authors estimate that geographical differences in cost of living are especially large for low-income US households, and that expensive cities offer college graduates a standard of living similar to affordable cities while offering less-educated households a lower one. A high school dropout household moving from the most affordable to the most expensive commuting zone would see consumption fall 18.5%.
  • Housing and the American Worker — Wages net of housing costs were once substantially higher in high-density areas than low-density areas for workers both with and without college education, but over the past half-century that urban wage premium has persisted only for college-educated workers, not for those with a high school degree or less.

AI Summary. The US Treasury aims to cut the federal deficit to 3% of GDP by 2028, paired with 3% sustained growth and an additional 3m barrels a day of oil production, arguing this and deregulation will lower long-term interest rates. The 10-year Treasury yield has traded between 4.3% and 4.7%, above the roughly 3.8% seen earlier.

Liz Capo McCormick, Enda Curran, and Gregory Korte Bloomberg
Date Posted:

At 4.3%, the 10-year is off recent highs of ~4.7%. “Not many on Wall Street are stating the case too loudly for Musk and Bessent engineering lower yields. There are small signs, though, that investors aren’t entirely ruling it out.”

Will deficit cuts and faster growth push long-term Treasury yields lower?

Core argument: The 10-year Treasury yield has swung between 4.3% and 4.7% since Trump took office, up from just under 4.3% on the eve of his election and around 3.8% in the summer, when Musk first floated the Department of Government Efficiency to Trump.

Since Trump took office, the 10-year yield has swung between 4.3% and 4.7%. This is up both from where it was on the eve of his election (just under 4.3%) and from back in the summer, when Musk first floated the idea to Trump of creating a group, the Department of Government Efficiency, to gut spending (around 3.8%). To try to push the rate down, Bessent’s goal is, with Musk’s help, to trim the deficit to 3% of GDP by 2028 and then to pair it with 3% sustained economic growth and boosting oil production by 3 million barrels a day. This, as well as deregulation, [Bessent] sees as set to trigger lower long-term rates. Not many on Wall Street are stating the case too loudly for Musk and Bessent engineering lower yields. There are small signs, though, that investors aren’t entirely ruling it out. US Treasuries in recent weeks have outperformed similar maturity interest-rate swaps. “The bond markets do not currently reflect the savings that I’m confident we can achieve,” [Musk said] “If you’re shorting bonds, I think you’re on the wrong side of the bet.”

Takeaways by Macro Roundup® AI

  1. The 10-year Treasury yield has swung between 4.3% and 4.7% since Trump took office, up from just under 4.3% on the eve of his election and around 3.8% in the summer, when Musk first floated the Department of Government Efficiency to Trump.
  2. Bessent's goal is to trim the deficit to 3% of GDP by 2028, paired with 3% sustained economic growth and oil production up 3 million barrels a day, which along with deregulation he sees as triggering lower long-term rates.
  3. Musk said the bond markets do not currently reflect the savings he is confident of achieving and that investors shorting bonds are on the wrong side of the bet, while US Treasuries have recently outperformed similar-maturity interest-rate swaps.

Related Articles:

  • Our Thoughts on Large US Deficits and Their Impact on Bond Yields — Total borrowing in the US is about average despite high government deficits, as high public borrowing has contributed to the need for higher interest rates that have depressed private sector borrowing. Risk premiums on long-dated debt look about average, with no market signs that economic or debt limits are currently being hit.
  • The Move in Long Rates Is Very Unusual — The Federal Reserve has cut its policy rate by 100 basis points while 10-year interest rates have risen by 100 basis points over the same stretch, a highly unusual divergence. Possible explanations include fiscal worries, weaker foreign demand for U.S. debt, or rate cuts that were not justified.
  • Data Update 4 for 2025: Interest Rates, Inflation and Central Banks! — Interest rates are ultimately driven by macroeconomic fundamentals, with an "intrinsic risk-free rate" defined as the sum of the inflation rate and real GDP growth tracking the observed 10-year Treasury bond yield closely. On this view, central banks' influence comes from preferential access to data on those fundamentals, their capacity to alter them, and their credibility.

AI Summary. A review of 695 hedge funds holding $3.1tn in gross equity positions ($2.0tn long, $1.1tn short) finds Amazon, Meta, Microsoft, Nvidia, Alphabet, and Apple remain the most popular long positions. Hedge funds trimmed the "Magnificent 7" on net, adding only to Tesla.

Ben Snider, Jenny Ma, David Kostin, Ryan Hammond, Daniel Chavez and Kartik Jayachandran Goldman Sachs
Date Posted:

Despite a modest drop in Magnificent 7 investments, Goldman Sachs finds that the most popular positions for 695 hedge funds with $3.1T of equity positions remain large-cap tech.

Are hedge funds still crowding into the largest technology stocks?

Core argument: The Hedge Fund Trend Monitor covers 695 hedge funds holding $3.1 trillion of gross equity positions, split between $2.0 trillion long and $1.1 trillion short.

This Hedge Fund Trend Monitor analyzes 695 hedge funds with $3.1 trillion of gross equity positions ($2.0 trillion long and $1.1 trillion short). The mega-caps continue to sit atop our list of the most popular hedge fund long positions. AMZN, META, MSFT, NVDA, GOOGL, and AAPL remain as the six most popular stocks. Hedge funds trimmed positions on net in the Magnificent 7 during 4Q. Except for TSLA, which funds added to during 4Q, the Magnificent 7 still sit atop of our Hedge Fund VIP list of the most popular long positions.

Takeaways by Macro Roundup® AI

  1. The Hedge Fund Trend Monitor covers 695 hedge funds holding $3.1 trillion of gross equity positions, split between $2.0 trillion long and $1.1 trillion short.
  2. AMZN, META, MSFT, NVDA, GOOGL, and AAPL remain the six most popular hedge fund long positions, with mega-caps atop the list.
  3. Hedge funds trimmed Magnificent 7 positions on net during 4Q, except for TSLA, which funds added to, and the Magnificent 7 still sit atop the Hedge Fund VIP list of most popular long positions.

Related Articles:

  • On Bubble Watch — Data from J.P. Morgan Asset Management covering monthly observations over a 27-year period show a strong relationship between the S&P 500's starting price-to-earnings ratio and its annualized return over the following ten years, with higher starting valuations consistently followed by lower returns. Purchases at a multiple near today's 22x were followed by ten-year returns between +2% and -2%.
  • US Exceptionalism: Drivers of Equity Outperformance and What’s Needed for a Repeat — US companies need about 7% earnings-per-share growth to deliver a normal 4% equity risk premium above the risk-free rate, versus about 3% for other developed markets, with the highest expectations embedded in technology and the largest US tech firms. Chinese companies could generate a 4% risk premium over Chinese bonds with no earnings growth.
  • Concentration & Correction – What To Do Next — The authors argue the equity market decline is a correction rather than the start of a sustained bear market, with valuations having reached very high levels, particularly in the US, where stocks remain expensive relative to history even excluding large-cap technology. They contend US market and technology-sector dominance reflects superior fundamentals rather than a bubble built on irrational exuberance.

AI Summary. Apple's pledge to spend more than $500bn in the U.S. over four years, including domestic suppliers and a facility making servers for its artificial-intelligence services, may largely reflect spending already expected. Wall Street projects nearly $1.3tn in total global spending over four years, and roughly 40% of that equals about $505bn.

Dan Gallagher Wall Street Journal
Date Posted:

Apple’s plans to spend $500B on capital investment and operating expenses in the US over the next four years reflect modest growth from the company’s 2021 to 2024 baseline.

Does a large U.S. investment pledge represent new spending?

Core argument: Apple announced plans to spend more than $500 billion in the U.S. over the next four years, including outlays with domestic suppliers and a manufacturing facility for servers supporting its Apple Intelligence service, plus a Detroit facility to train U.S. manufacturers.

Apple announced plans Monday morning to spend more than $500 billion in the U.S. over the next four years. The outlay will include money spent with domestic suppliers and on the opening of a manufacturing facility to produce the servers needed to support the company’s artificial-intelligence service, called Apple Intelligence. The company also said it plans to open a facility in Detroit to train “the next generation of U.S. manufacturers.” Unclear, though, is how much of the planned spending is actually new. Apple has spent about $1.1 trillion over the past four fiscal years on total operating expenses and capital expenditures—and Wall Street expects nearly $1.3 trillion in total spending over the next four years, according to consensus estimates by Visible Alpha. While Apple doesn’t break out its expenses per geography, about 43% of its revenue comes from the Americas region, which it defines as North and South America. Assuming the U.S. constitutes the large bulk of that number, and if spending is about in line with revenue, then a rough figure of 40% of projected global spending through the 2028 fiscal year equates to about $505 billion.

Takeaways by Macro Roundup® AI

  1. Apple announced plans to spend more than $500 billion in the U.S. over the next four years, including outlays with domestic suppliers and a manufacturing facility for servers supporting its Apple Intelligence service, plus a Detroit facility to train U.S. manufacturers.
  2. Wall Street consensus estimates from Visible Alpha put Apple's total global spending at nearly $1.3 trillion over the next four years, compared with about $1.1 trillion in operating expenses and capital expenditures over the past four fiscal years.
  3. About 43% of Apple's revenue comes from the Americas, and applying a rough 40% share to projected global spending through fiscal 2028 yields roughly $505 billion, leaving unclear how much of the pledged U.S. investment is new.

Related Articles:

  • Microsoft Dropped Some AI Data Center Leases, TD Cowen Says — Microsoft has voided data center leases totaling a couple of hundred megawatts of capacity and stopped converting preliminary agreements into formal leases, according to brokerage TD Cowen's checks with supply chain providers. TD Cowen also says Microsoft is redirecting some planned international spending to the US, suggesting a material slowdown in international leasing.
  • A Severe Case of COVIDIA: Prognosis For An AI-Driven US Equity Market — Computing cycles move from infrastructure to platforms to applications, and NVIDIA data center revenues as a share of market-wide capital spending are projected to reach levels last seen at the 1969 mainframe peak and the dot-com boom. Large cloud providers spend more than twice as much training foundational models as running them for customers.
  • Are We on the Brink of an AI Investment Arms Race? — Spending on AI chips and data centers was a bit below $100bn in 2023, around 0.4% of US GDP, and the broader set of companies exposed to AI-related investment has not yet seen a major jump in sales. The authors argue the underlying dynamics point to a spending surge large enough to shape the business cycle.

AI Summary. China's leadership held a high-level symposium with private sector entrepreneurs, including leaders in technology and emerging industries, which can produce short-term boosts to confidence, household consumption, and business activity. The author argues that without shifting the growth model toward a higher household share of GDP—requiring transfers from businesses or governments—such confidence-boosting will not sustainably raise domestic demand.

Michael Pettis China Financial Markets
Date Posted:

Xi’s belated public embrace of China’s entrepreneurs won’t restore robust growth until policy changes to raise household spending and domestic demand. @michaelxpettis

Can boosting private sector confidence sustainably raise China's domestic demand?

Core argument: Xi Jinping spoke at a high-level symposium of private sector entrepreneurs on February 17, the first meeting of its kind since 2018, attended by leading figures from China's technology and emerging sectors.

On February 17, Chinese President Xi Jinping set off a frisson of economic excitement by speaking at a high-level symposium of private sector entrepreneurs. The first meeting of its kind since 2018, it was attended by some of the leading lights of China’s technology and emerging sectors. The right kind of pep talks may indeed create short-term boosts to confidence, which in turn can create short-term boosts to household consumption and spur businesses to eagerly respond. However, the structural view of the decline in consumption growth implies that without a transformation of the growth model toward a higher household share of GDP—which inevitably requires either explicit or implicit transfers from businesses or governments, showing why it is so difficult to pull off)—confidence-boosting pep talks won’t create sustainable increases in domestic demand.

Takeaways by Macro Roundup® AI

  1. Xi Jinping spoke at a high-level symposium of private sector entrepreneurs on February 17, the first meeting of its kind since 2018, attended by leading figures from China's technology and emerging sectors.
  2. The right kind of pep talks may create short-term boosts to confidence, which can in turn lift household consumption and spur businesses to respond.
  3. The structural view of China's declining consumption growth implies that without shifting the growth model toward a higher household share of GDP—which requires explicit or implicit transfers from businesses or governments—confidence-boosting talks won't sustainably raise domestic demand.

Related Articles:

  • 2023 Letter — Rising numbers of Chinese nationals are being apprehended at the US-Mexico border, with US officers encountering around 1,000 in one month early in the year and nearly 5,000 by November. Most Chinese emigrants leave through legal means, going primarily to Singapore, Japan, and Thailand.
  • Chinese Venture Capitalists Force Failed Founders On To Debtor Blacklist — More than 80% of venture and private equity deals in China contain redemption clauses requiring companies, and often founders personally, to buy back investors' shares with interest if targets such as an initial public offering timeline or revenue goals are missed, according to estimates by law firm Lifeng Partners.
  • DeepSeek FAQ — Consumers and businesses stand to gain most from a future of effectively free AI products, along with big consumer tech companies, since a world of free AI makes product and distribution the decisive advantages they already hold. China also benefits, and the United States faces a choice between expanding chip export restrictions and permission-based regulation or allowing itself to compete.