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
  • “…challenges misconceptions that distort our economic debates.” - Arthur Brooks, President of the American Enterprise Institute
  • “…a must-read for serious students of economic policy.” - Glenn Hubbard, Dean, Columbia Business School, and former Chairman of the Council of Economic Advisers
  • “…a must-read for serious students of economic policy.” - Glenn Hubbard, Dean, Columbia Business School, and former Chairman of the Council of Economic Advisers
  • “…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
  • “…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
  • “Unintended Consequences represents the most cogent and persuasive analysis of the Financial Crisis to date.” - Andrei Shleifer, 1999 John Bates Clark Medal Winner
  • “Unintended Consequences is full of substance, it is one of the must-read books of the year, and once I finish it I will be giving it a second read through right away.” - Tyler Cowen, Professor, George Mason University
  • “…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
  • “Unintended Consequences is far smarter and more thought-provoking than most economics written for the general public” - Greg Mankiw, Harvard University, Former Chairman of the Council of Economic Advisors
  • “Unintended Consequences should be read by anyone who takes for granted the superiority of progressive taxation and has not thought carefully about the trade-offs involved.” - The New Republic
Upside of Inequality Oxford Unintended Consequences

Macro Roundup


Friday, November 22, 2024

AI Summary. Passive funds hold about 50% of global equity mutual fund and exchange-traded fund assets, and roughly 60% of US equities. The authors argue that as active investors shift to passive strategies, fewer short sellers remain, raising large-cap volatility and putting upward pressure on large-cap prices.

Felix von Moltke and Torsten Sløk Apollo
Date Posted:

Torsten Sløk argues the rise of passive investing (~60% of US equity funds) has led to reduced shorting, increasing market volatility, reducing liquidity, and rising P/E ratios for large-cap stocks like the “Magnificent Seven.”

How does the shift to passive investing affect large-cap volatility and prices?

Core argument: Passive ownership of global equity mutual funds and ETFs stands at 50%, reaching 60% for US equities and close to 50% for non-domestic equities, with flows into passive equity growing at a measured pace since 2012.

We define a passive investor as one who always holds every security in the same manner as in the market [effectively] one who invests in index funds or passive ETFs. Using that definition, passive ownership of global equity mutual funds and ETFs is 50%. While the rate of flows into passive equity has grown at a measured pace since 2012, passive equity commands a sizable share of AUM - 60% for US equities, with non-domestic equities close to 50%. Passive investors are always long. If more active investors turn passive, then there will be fewer investors who are short, which again will increase the volatility of large cap stocks. Because large cap stocks have a bigger proportional impact on portfolios, the higher volatility of large cap stocks will put upward pressure on large cap stock prices. In short, when active investors turn passive, large cap stocks will benefit disproportionately. This dynamic can be observed in the price-to-earnings (P/E) ratios of large- and mega-cap stocks, which have been consistently high and growing, especially for the Magnificent Seven.

Takeaways by Macro Roundup® AI

  1. Passive ownership of global equity mutual funds and ETFs stands at 50%, reaching 60% for US equities and close to 50% for non-domestic equities, with flows into passive equity growing at a measured pace since 2012.
  2. The authors argue that because passive investors are always long, a shift from active to passive leaves fewer short investors and raises the volatility of large cap stocks.
  3. The authors argue that higher large cap volatility puts upward pressure on large cap prices, a dynamic they observe in the consistently high and growing P/E ratios of large- and mega-cap stocks, especially the Magnificent Seven.

Related Articles:

  • The Less-Efficient Market Hypothesis — Markets are never perfectly efficient because betting against other investors' errors carries risk, and the author argues social media, investor overconfidence, and gamified 24/7 trading have made markets meaningfully less efficient, with indexing and very low interest rates a smaller contributor. The recommended response is to adopt the longest possible investment time horizon.
  • Updating Our Long-Term Return Forecast For US Equities to Incorporate the Current High Level of Market Concentration — The authors estimate the S&P 500 will deliver an annualized nominal total return of 3% over the next 10 years, roughly 1% in real terms, modeling long-term returns as a function of starting valuation, market concentration, frequency of economic contractions, corporate profitability, and interest rates. The index returned 13% annualized over the past decade, above its 11% long-term average.
  • Stock Market Concentration — Over 2014–2023 the 10 largest U.S. stocks averaged 19% of market capitalization while generating 47% of total economic profit, and by 2023 they were 27% of market capitalization and 69% of economic profit. The S&P 500 returned above its average when concentration was rising and below it when concentration was falling.

AI Summary. The authors estimate that net international migration to the US was close to 3.5m in one year — about 200,000 above the Congressional Budget Office's estimate, driven largely by higher undocumented entries — and project roughly 2.9m the following year, about 400,000 below the Congressional Budget Office's projection.

Evgeniya Duzhak Federal Reserve Bank of San Francisco
Date Posted:

US net international migration was almost 3.5mm in 2023 and is on track to decline by ~ 500k to 3.0mm in 2024. Inflows are 3x the historical average. The inflow of undocumented immigrants rose throughout 2023, to a peak of ~ 1.1 mm in the 4th quarter.

Is net immigration running higher than official estimates and slowing faster?

Core argument: The authors estimate net international migration to the U.S. was close to 3.5 million in 2023, about 200,000 higher than the CBO's prior estimate, with the gap largely driven by higher numbers of undocumented immigrants.

We use the most recent data from the Department of Homeland Security through September 2024 to further update the number of undocumented immigrant entries and estimates of NIM in 2023 and 2024. We estimate that NIM in 2023 was close to 3.5 million, about 200,000 higher than previously estimated by the CBO. Furthermore, new data suggest that the sharp increase in NIM in 2023 may be subsiding more quickly than previously estimated by the CBO in January 2024. Incorporating this new data and projecting to the end of 2024 suggests that NIM levels in 2024 may be around 2.9 million, about 400,000 lower than the CBO’s projection. The increase in NIM compared with the CBO’s estimate for 2023 is largely driven by higher numbers of immigrants in the undocumented category. Figure 1 shows how the inflow of undocumented immigrants increased throughout 2023 to a peak of around 1.1 million in the fourth quarter.

Takeaways by Macro Roundup® AI

  1. The authors estimate net international migration to the U.S. was close to 3.5 million in 2023, about 200,000 higher than the CBO's prior estimate, with the gap largely driven by higher numbers of undocumented immigrants.
  2. Projecting Department of Homeland Security data through September 2024 to year-end, the authors estimate 2024 net international migration of around 2.9 million, about 400,000 below the CBO's January 2024 projection.
  3. Inflows of undocumented immigrants rose through 2023 to a peak of roughly 1.1 million in the fourth quarter, and new data suggest the 2023 surge in net international migration may be subsiding more quickly than the CBO estimated.

Related Articles:

  • The Demographic Outlook: 2024 to 2054 — The Congressional Budget Office projects the US population will grow from 342mm in 2024 to 383mm in 2054, an average 0.4% a year — less than half the 0.9% pace of the prior five decades. Immigration accounts for about 70% of growth over the next decade and for all population growth beginning in 2040.
  • Immigration and the Macroeconomy After 2024 — Projected net migration ranges from 3.7mm under a high-immigration policy scenario to negative 740,000 under a low-immigration scenario, implying four-year growth in the working-age civilian population of anywhere from 5mm to 15mm people. The authors project the resulting gap in annual US economic growth at roughly half a percentage point, or about $130bn.
  • The Cost of Illegal Immigration — The US foreign-born population rose by 6.6mm to 51.6mm over a roughly three-year span, lifting immigrants to 15.6% of the population, above the prior records of 14.8% in 1890 and 14.7% in 1910. The authors estimate about 13.7mm illegal immigrants now live in the US, a net increase of 3.7mm.

AI Summary. In a randomized clinical trial, physicians given access to a large language model scored a median 76% on diagnostic reasoning cases versus 74% for those without it, a 2-percentage-point difference that was not statistically significant. A sensitivity analysis including incomplete cases produced the same 2-point difference.

Ethan Goh, Robert Gallo, and Jason Horn Journal of the American Medical Association
Date Posted:

In a randomized clinical trial of diagnostic reasoning, physicians using conventional resources scored 74%, while those using GPT-4 scored 76%. GPT-4 alone scored 16pp higher than the conventional resources group.

Does giving physicians a large language model improve diagnostic reasoning?

Core argument: Physicians randomized to use a large language model scored a median 76% (IQR, 66%-87%) per case on diagnostic reasoning versus 74% (IQR, 63%-84%) for controls, a difference of 2 percentage points (95% CI, −4 to 8; P = .60).

A total of 244 cases were completed by all participants (125 cases in LLM group, 119 cases in control group). The median number of completed cases per participant was 5 (IQR, 4-6). Analysis of the transcripts showed that 100% (22 of 22) of physicians randomized to use the LLM did so; 3 transcripts were lost due to technical issues and not included. The median score per case was 76% (IQR, 66%-87%) for the LLM group and 74% (IQR, 63%-84%) for the control group. The mixed-effects model showed a difference of 2 percentage points (95% CI, −4 to 8 percentage points; P = .60) between the LLM and control groups, as presented in Table 2. A sensitivity analysis including all cases, complete and incomplete, showed a similar result with a difference of 2 percentage points (95% CI, −4 to 8 percentage points; P = .50) between the LLM and control group.

Takeaways by Macro Roundup® AI

  1. Physicians randomized to use a large language model scored a median 76% (IQR, 66%-87%) per case on diagnostic reasoning versus 74% (IQR, 63%-84%) for controls, a difference of 2 percentage points (95% CI, −4 to 8; P = .60).
  2. A sensitivity analysis including complete and incomplete cases showed a similar 2-percentage-point difference between the large language model and control groups (95% CI, −4 to 8; P = .50).
  3. All 22 physicians randomized to the large language model used it, with 244 cases completed across groups (125 in the LLM group, 119 in the control group) and a median of 5 cases per participant (IQR, 4-6).

Related Articles:

  • AI-Enabled Electrocardiography Alert Intervention and All-Cause Mortality: a Pragmatic Randomized Clinical Trial — In a randomized trial of 15,965 hospitalized patients, an artificial-intelligence electrocardiogram system that flagged patients predicted to be at high risk of death and sent warning messages to physicians was associated with lower 90-day all-cause mortality, 3.6% in the intervention group versus 4.3% in the control group.
  • Assessing the Implications of a Productivity Miracle — If artificial intelligence can perform economically valuable cognitive tasks at very low marginal cost, consumers would buy far more of the cheapened goods and spend the savings in other categories, expanding activity and employment in sectors facing new demand. Where supply is inherently limited, such as real estate, those savings could bid up prices and produce inflation.

AI Summary. The US federal government employs 4.3mm people — 3mm civilians and 1.3mm active-duty military — with about 2.4mm civilian jobs outside the Postal Service. Those non-Postal civilian workers account for 1.5% of nonfarm payroll employment, down from over 3% in the 1950s and 1960s, while state and local government jobs make up nearly 13%.

Justin Fox Bloomberg
Date Posted:

There are 4.3mm Federal employees, 3mm civilian, and 1.3mm active duty military. Excluding postal service workers, civilian federal workers make up 1.5% of nonfarm payroll employment, down from 2% in the early 1990s. @foxjust

How large is the federal workforce relative to total US employment?

Core argument: Non-Postal Service civilian federal employment stood at 2,395,900 as of October, according to the US Bureau of Labor Statistics, out of 4.3mm total federal employees comprising 3mm civilians and 1.3mm active-duty military.

There are 4.3mm federal government employees — 3mm civilians and 1.3mm active-duty military. It’s non-Postal Service civilian jobs that the prospective Trump administration seems to have in its crosshairs, and there were 2,395,900 of those as of October, according to the US Bureau of Labor Statistics. This number has grown over the past 25 years, including an increase of 76,300 during Trump’s first term, but the growth followed sharp declines in the mid-1990s. Current non-Postal Service federal employment is slightly higher than in the late 1980s and early 1990s. Include the Postal Service, and it’s slightly lower. Civilian federal workers (Postal Service excluded) now make up just 1.5% of nonfarm payroll employment, down from more than 3% in the 1950s and 1960s and more than 2% through the early 1990s. State and local government jobs together make up almost 13% of nonfarm payroll

Takeaways by Macro Roundup® AI

  1. Non-Postal Service civilian federal employment stood at 2,395,900 as of October, according to the US Bureau of Labor Statistics, out of 4.3mm total federal employees comprising 3mm civilians and 1.3mm active-duty military.
  2. Non-Postal Service federal employment has grown over the past 25 years, including an increase of 76,300 during Trump's first term, but that growth followed sharp declines in the mid-1990s, leaving the level slightly above that of the late 1980s and early 1990s.
  3. Civilian federal workers excluding the Postal Service make up 1.5% of nonfarm payroll employment, down from more than 3% in the 1950s and 1960s, while state and local government jobs together account for almost 13%.

Related Articles:

  • Interest Expense: A Bigger Impact on Deficits than Debt — The US is running a primary deficit—excluding interest costs—much larger than its historical norm, at a point in the business cycle when deficits would normally be smaller than usual. The average interest rate on federal debt is projected to remain at or below the rate of nominal GDP growth over the next decade.
  • Why Trump’s Trade War Will Cause Chaos — The trade balance reflects the gap between national income and spending, so it will not change unless domestic saving and investment do. With foreign savings inflows averaging 3.9% of GDP while households and corporations ran savings surpluses, the author argues eliminating the external deficit requires shrinking the government's 6.7%-of-GDP deficit.
  • 2023 Chart Book Examines Spending, Taxes, and Deficits — The six largest US deficit reduction deals since 1983 were scored at enactment as saving between 0.52% and 1.72% of GDP, with the 1985 Gramm-Rudman-Hollings Act the largest. Many of the cuts were later reversed, and the 1985 law was invalidated by the Supreme Court and replaced in 1987.

AI Summary. Four Democratic Senate candidates won in states that Trump flipped from Democratic to Republican, an outcome that has occurred only twice between 2004 and 2020, since the party winning a flipped state almost always wins its concurrent Senate race. Ticket-splitting, voters who backed Trump but left the rest of the ballot blank, and minor third-party Senate votes all contributed.

Nate Moore The Liberal Patriot
Date Posted:

.@NateMoore1287 notes that Democrats retained Senate seats in 4 out of the 5 swing states Trump flipped in 2024. While it’s hard to disentangle ticket-splitting from an undervote, this raises questions about the sustainability of Trump’s 2024 coalition.

How did Democratic Senate candidates win states the presidential nominee flipped?

Core argument: Four Democratic senators won in states Trump flipped from blue to red, an outcome that defies recent history in which the party flipping a presidential state almost always wins the concurrent Senate race.

Four Democrats won in states that Trump flipped from blue to red. Such a feat defies recent political history. When a presidential candidate flips a state, their party almost always wins the concurrent Senate race; e.g. Democrats won all Senate races in states that Biden flipped blue in 2020. Republicans did the same in 2016. In fact, between 2004 and 2020 such split outcomes have happened just twice—Democrat Joe Donnelly’s 2012 win in Indiana and Republican Judd Gregg’s 2004 victory in New Hampshire. How did Democrats eke out Senate wins in states that Trump flipped? Each certainly benefited from ticket-splitting. But Democratic success may also owe to the traditional “undervote”—voters who selected Trump and left the rest of their ballot blank. Others, meanwhile, voted for a minor third-party Senate candidate, effectively the same as an undervote. The undervote, of course, is difficult to disentangle from persuasion effects. What share of Trump-only voters will show up in the midterms? Can the GOP retain their first-time non-white voters? Realignment—or not—hangs in the balance.

Takeaways by Macro Roundup® AI

  1. Four Democratic senators won in states Trump flipped from blue to red, an outcome that defies recent history in which the party flipping a presidential state almost always wins the concurrent Senate race.
  2. Between 2004 and 2020, split presidential-Senate outcomes in flipped states occurred just twice: Democrat Joe Donnelly's 2012 win in Indiana and Republican Judd Gregg's 2004 victory in New Hampshire, while Democrats won every Senate race in states Biden flipped in 2020 and Republicans did the same in 2016.
  3. Democratic Senate wins may owe to ticket-splitting, to an "undervote" of voters who chose Trump and left the rest of the ballot blank, and to voters backing minor third-party Senate candidates, effects difficult to disentangle from persuasion.

Related Articles:

  • The Shattering of the Democratic Coalition — The Democratic presidential margin among women narrowed from 12 points to 7 while the Republican margin among men widened from 5 points to 10, leaving the overall gender gap unchanged at 17 points. Women under 30 shifted 14 points toward Republicans and men under 30 shifted 29 points.
  • “Compensate the Losers?” Economic Policy and Partisan Realignment in the US — The authors argue that partisan realignment by education reflects Democrats' declining supply of "predistribution" policies such as job guarantees, higher minimum wages, protectionism, and stronger unions, which less-educated Americans favor while more-educated Americans prefer taxes and transfers. They estimate changing views of the parties' economic policies explain roughly half the shift.
  • Key to Trump’s Win: Heavy Losses for Harris Across the Map — The Republican presidential candidate added about 2.5m votes relative to the prior election, while the Democratic candidate received about 7m fewer votes than her party's previous nominee. An analysis found she failed to add voters in three of the seven swing states and in 80% of US counties.

AI Summary. A serologic survey of 115 dairy workers in Michigan and Colorado found that 8 (7%) had evidence of recent infection with highly pathogenic H5 bird flu virus, all of whom reported milking cows or cleaning the milking parlor. Four of those infected recalled illness around the time cows were sick.

Alexandra Mellis, Joseph Coyle, Kristen Marshall et al. Centers for Disease Control and Prevention
Date Posted:

A @CDCgov serologic survey of dairy workers found 7% (8 of the 115) had serologic evidence of recent infection with the A(H5) bird flu virus. The virus may be widespread in human populations already.

Are dairy workers being infected with bird flu without detection?

Core argument: Among 115 dairy workers surveyed in Michigan and Colorado during June–August 2024, eight (7%; 95% CI = 3.6%–13.1%) had serologic evidence of recent infection with highly pathogenic avian influenza A(H5) virus.

During June–August 2024, CDC collaborated with the Michigan Department of Health and Human Services and the Colorado Department of Public Health and Environment to implement cross-sectional serologic surveys to ascertain the prevalence of recent infection with HPAI A(H5) virus among dairy workers. In both states, a convenience sample of persons who work in dairies was interviewed, and blood specimens were collected. Among 115 persons, eight (7%; 95% CI = 3.6% – 13.1%) had serologic evidence of recent infection with A(H5) virus; all reported milking cows or cleaning the milking parlor. Among persons with serologic evidence of infection, four recalled being ill around the time cows were ill; symptoms began before or within a few days of A(H5) virus detections among cows.

Takeaways by Macro Roundup® AI

  1. Among 115 dairy workers surveyed in Michigan and Colorado during June–August 2024, eight (7%; 95% CI = 3.6%–13.1%) had serologic evidence of recent infection with highly pathogenic avian influenza A(H5) virus.
  2. All eight dairy workers with serologic evidence of recent A(H5) infection reported milking cows or cleaning the milking parlor, according to CDC surveys conducted with Michigan and Colorado health agencies.
  3. Four of the workers with serologic evidence of A(H5) infection recalled illness around the time cows were ill, with symptoms beginning before or within a few days of A(H5) virus detections among cows.

Related Articles:

  • US Spots First H5N1 Bird Flu Case in a Pig, Raising Concern for Humans — A pig on a backyard farm in Oregon tested positive for H5N1 bird flu, the first such case in a US pig, with infection found throughout its body despite no signs of illness. Pigs are considered mixing vessels in which human and bird flu viruses can exchange genes, raising concern that hybrid strains could emerge that infect humans more easily.
  • Possible Cluster of Human Bird-Flu Infections Expands in Missouri — The Centers for Disease Control and Prevention reported a possible cluster of eight bird-flu cases in Missouri, which could represent the first person-to-person transmission of the H5N1 virus in the United States. Infections in seven of the individuals, a household member and six healthcare workers, have not been confirmed, leaving open other flulike illnesses.
  • How the Current Bird Flu Strain Evolved to be so Deadly — A Nature study tracking bird flu genome changes and reported outbreaks found the virus spread three times faster among wild birds than farmed poultry, attributed to mutations letting it adapt to diverse species. Outbreaks, once seasonal with Northern Hemisphere bird migration, have become persistent, killing millions of birds across five continents.

Monday, November 25, 2024

AI Summary. The authors estimate that men born in US counties more exposed to textile-industry decline lost about 3.3 months of old-age longevity if they lived in non-urban areas and did not migrate, with effects concentrated where alternative job prospects were fewer. Scaling by the employment drop implies roughly 4 years lost among sons of displaced textile workers.

Hamid Noghanibehambari and Jason Fletcher National Bureau of Economic Research
Date Posted:

Hamid Noghanibehambari and @jasonmfletcher find that children whose fathers lost their jobs in textiles and did not migrate during the decline of the New England textile industry in the 1920s and 1930s experienced “a 4-year loss of life expectancy.”

Does local industrial decline in childhood shorten men's later-life lifespans?

Core argument: Men born in New England counties more exposed to post-1920s textile deindustrialization show reductions in old-age longevity of about 3.3 months if they lived in non-urban areas and their families had not migrated from their county of birth by 1940.

We compare old-age longevity of individuals born in counties with higher versus lower exposure to the deindustrialization after the 1920s versus before and find reductions in longevity of about 3.3 months for exposed cohorts who reside in non-urban areas and whose family did not migrate from their county-of-birth until 1940. The negative impacts are primarily concentrated among people and places with fewer alternative job prospects. These effects measure the potential impacts among all people in as defined above. We expect to observe larger effects as we focus on the subpopulations who are more likely to be affected- [e.g.] low educated individuals [who] are less likely to switch occupations.We find that high-exposed counties (top-tercile of 1900 textile) reveal about 7 percentage-points drop in textile employment. Using the 3.3 months applying to the [at risk cohorts as described above] and deflating by the first-stage effect of 0.07 in Table 4 [“father not in textile industry”], a back-of-an-envelope calculation suggests a loss [of life expectancy] about 4 years among non-urban non-migrants whose fathers lost their job due to deindustrialization.

Takeaways by Macro Roundup® AI

  1. Men born in New England counties more exposed to post-1920s textile deindustrialization show reductions in old-age longevity of about 3.3 months if they lived in non-urban areas and their families had not migrated from their county of birth by 1940.
  2. Counties in the top tercile of 1900 textile employment show about a 7 percentage-point drop in textile employment, and scaling the 3.3-month effect by that first stage suggests a loss of roughly 4 years of life expectancy among non-urban non-migrants whose fathers lost textile jobs.
  3. The negative longevity effects are concentrated among people and places with fewer alternative job prospects, with larger effects expected among subpopulations such as low-educated individuals, who are less likely to switch occupations.

Related Articles:

  • The China Shock Revisited: Job Reallocation and Industry Switching in U.S. Labor Markets — Using U.S. Census Bureau administrative data, the authors find that rising Chinese import penetration shifted jobs from manufacturing to services far more in high human capital areas, where within-firm industry switching occurs almost exclusively, producing a net positive employment effect; in low human capital, manufacturing-dependent areas, service gains barely offset manufacturing losses.
  • Trading Places: Mobility Responses of Native and Foreign-Born Adults to the China Trade Shock — U.S. local labor markets more exposed to import competition from China saw substantially larger net reductions in their foreign-born populations, with no significant reduction among native-born residents. Within exposed areas, foreign-born and native-born workers had comparably sized declines in employment-to-population ratios.
  • Manufacturing Jobs Boom Not Reaching Places Hit By The China Shock — U.S. manufacturing job growth since the pandemic is occurring disproportionately in counties least exposed to Chinese import competition in the 2000s, according to an analysis of federal employment and wage data. The counties hardest hit by that import competition still have fewer manufacturing jobs than before the pandemic.

AI Summary. US federal debt has risen from 105% to 123% of GDP, the budget deficit has doubled from 3.1% to more than 6%, and government interest expense has grown from 1.3% to about 3.1% of GDP. The author argues this makes expansive fiscal policy harder to pursue than in the past.

Jens Nordvig Money: Inside and Out
Date Posted:

.@jnordvig notes that Trump ran an expansionary fiscal policy in his first term, growing the deficit 1.5pp to 4.6% of GDP in 2019. Scott Bessent’s goal of halving the deficit from 6.4% of GDP in 2024 to 3% by 2029 implies a contractionary policy.

Do higher debt, deficits, and interest costs constrain expansive fiscal policy?

Core argument: The U.S. federal deficit doubled from 3.1% of GDP in 2016 to more than 6% in 2024, while federal debt rose from 105% to 123% of GDP, which will make expansive fiscal policy harder to pursue than in Trump's first term.

We also saw notable expenditure expansion during Trump I, leading to a significant expansion in the fiscal deficit, even before the COVID shock hit in 2020. Specifically, the fiscal deficit was 3.1% in 2016, but had grown to 4.6% in 2019. This may seem like a small deterioration, and visually it is small compared to the mega-deficit of 2020. But historically it was a quick 1.5% deterioration which was rare in an economy experiencing ‘ok growth.’ But 2024 is different from 2016. The Federal debt level has grown from 105% of GDP to 123% of GDP. And the fiscal deficit has doubled, from 3.1% in 2016, to more than 6% in 2024. Hence, it will be hard to pursue expansive fiscal policies in the same way as in the first Trump administration. Linked to this, we have seen much more volatility in US bond markets, with supply concerns contributing to yield spikes during 2022 and 2023, and perhaps recently too. The 10-year government bond yield is currently around 4.50% compared to less than 2% when Trump was elected in 2016, with important implications for the debt service bill and debt dynamics (the government interest expense has grown from 1.3% in 2016 to around 3.1% in 2024).

Takeaways by Macro Roundup® AI

  1. The U.S. federal deficit doubled from 3.1% of GDP in 2016 to more than 6% in 2024, while federal debt rose from 105% to 123% of GDP, which will make expansive fiscal policy harder to pursue than in Trump's first term.
  2. During Trump's first term, the federal deficit widened from 3.1% of GDP in 2016 to 4.6% in 2019, a 1.5 point deterioration that was historically rare in an economy experiencing "ok growth," and before the 2020 COVID shock.
  3. The 10-year government bond yield stands near 4.50% versus less than 2% when Trump was elected in 2016, and government interest expense has grown from 1.3% in 2016 to around 3.1% in 2024.

Related Articles:

  • Long-Term Interest Rates, Fiscal Policy, And The Term Premium — US fiscal policy is on an unsustainable path, with potential market flashpoints including a sovereign credit downgrade, weak Treasury auctions, a term premium that has been trending higher, and a weaker dollar if growth slows. A debt crisis is unlikely; gradually higher long-term interest rates and a steeper yield curve are more likely.
  • As the Fed Cuts Rates, Treasury Yields Are Rising — Federal Reserve cuts to the federal funds rate target range were accompanied by falling short-term Treasury bill yields but rising longer-term yields, with the 10-year Treasury yield climbing from 3.63% to 4.42%. Deficit-increasing proposals from both presidential campaigns likely exacerbated the rise in longer-term yields.
  • How to Manage the Dollar (Part III) — Policy options for influencing the dollar fall into four overlapping categories: acting on the current account, acting on the financial account, monetary policy, and coordinated global policies affecting saving-investment balances. Because the US current account is driven by the economy's overall saving-investment balance, any policy affects the external imbalance only by shifting those decisions.

AI Summary. US economic output per hour worked has risen 8.9% over the past five years, faster than the preceding five years or any point in the 2010s, and more than double the growth of the next-fastest major comparable economy. Unlike the second-place UK, the US achieved this while also increasing overall employment.

Joseph Politano Apricitas Economics
Date Posted:

American economic output per hour worked has risen 8.9% since 2019—faster than any point in the 2010s —in spite of the COVID-19 pandemic, and more than double the productivity growth of the UK, the next-fastest major comparable economy.

Is US productivity growing faster than in other major economies?

Core argument: U.S. output per hour worked has risen 8.9% over the last five years, faster than the prior five years or any point in the 2010s, despite the COVID-19 pandemic.

American economic output per hour worked has risen 8.9% over the last five years—faster than the five years prior or any point in the 2010s—in spite of the COVID-19 pandemic. The already-great productivity recovery we had going into this year got a major upward revision from recent updates to GDP data, and preliminary estimates suggest they’ll get another boost as job growth will likely be revised down significantly at the start of next year. That success is all the more remarkable compared to the dismal productivity numbers seen across many of America’s peer nations and how long it took US productivity growth to sustainably recover from the Great Recession. Since late 2019, the US has seen more than double the productivity growth of the next-fastest major comparable economy, building on an already significant lead it built up in the years before COVID. Plus, unlike the 2nd-place UK, the US has achieved this while increasing overall employment levels instead of leaving less-productive workers out of a job.

Takeaways by Macro Roundup® AI

  1. U.S. output per hour worked has risen 8.9% over the last five years, faster than the prior five years or any point in the 2010s, despite the COVID-19 pandemic.
  2. Since late 2019, the U.S. has recorded more than double the productivity growth of the next-fastest major comparable economy, building on a lead established before COVID.
  3. Unlike the second-place UK, the U.S. achieved its productivity gains while increasing overall employment rather than leaving less-productive workers out of a job.

Related Articles:

  • American Productivity Still Leads The World — US labor productivity per hour has grown 73% since 1990, versus 39% in the euro area and 55% in both Britain and Japan. America also invests more, with non-residential investment near 17% of GDP and R&D around 3.5%, and sees company churn of nearly 20% annually versus roughly 15% in Europe.
  • Does Worker Scarcity Spur Investment, Automation and Productivity? Evidence from Earnings Calls — The authors estimate that a one-unit increase in a firm's labor issues, as measured by earnings calls, raises its investment by 28 basis points, implying roughly $55bn in additional US investment attributable to tighter labor markets. The effect is concentrated in industries relying heavily on routine manual tasks, such as manufacturing assembly and warehouse packaging.
  • Why is the U.S. GDP Recovering Faster Than Other Advanced Economies? — The U.S. output advantage over other advanced economies is attributed mainly to faster labor productivity growth and secondarily to hours worked per employee. Rigid foreign labor markets and job-retention policies kept workers attached to firms producing below capacity, while the U.S. saw about 3x more sectoral worker reallocation and a surge in new firm formation.

AI Summary. Private equity funds managed by Goldman Sachs will write off nearly $900mn of exposure to Swedish battery maker Northvolt, taking their stake to zero after the company filed for Chapter 11 bankruptcy. The bank had been Northvolt's second-largest shareholder, and one fund had recently valued its stake at 4.29x cost.

Richard Milne and Harriet Agnew Financial Times
Date Posted:

Goldman Sachs is writing its $896mm investment in Swedish battery maker Northvolt to zero. As recently as 7 months ago the position was marked at a 4.29x multiple-of-money, and they had projected an increase to a 6x multiple over the next year.

Do private equity valuations overstate the worth of failing startups?

Core argument: Goldman Sachs private equity funds will write off at least $896mn of exposure after Swedish battery maker Northvolt filed for Chapter 11 bankruptcy, marking the stake down to zero at year-end.

Funds managed by Goldman Sachs will write off almost $900mn after Swedish battery maker Northvolt filed for Chapter 11 bankruptcy. Goldman’s private equity funds have at least $896mn in exposure to Northvolt, making the US bank its second-largest shareholder. They will write that down to zero at the end of the year. The losses mark a sharp contrast to a bullish prediction just seven months ago by one of the Goldman funds, which told investors that its investment in Northvolt was worth 4.29 times what it had paid for it, and that this would increase to six times by next year.

Takeaways by Macro Roundup® AI

  1. Goldman Sachs private equity funds will write off at least $896mn of exposure after Swedish battery maker Northvolt filed for Chapter 11 bankruptcy, marking the stake down to zero at year-end.
  2. Goldman Sachs is Northvolt's second-largest shareholder through its private equity funds.
  3. Seven months before the write-off, one Goldman Sachs fund told investors its Northvolt investment was worth 4.29 times what it had paid, and projected that multiple would reach six times by next year.

Related Articles:

  • The Climate Short: Hedge Funds Pile Up Huge Bets Against Green Future — An analysis of positions voluntarily disclosed by roughly 500 hedge funds found more funds are net short batteries, solar, electric vehicles and hydrogen than long them, while more are net long fossil fuels than short oil, gas and coal. This comes despite large green stimulus programs in the US, Europe and China.
  • It’s Mostly a Paper Moon: Alternative Investments Review — Venture capital financing has tightened: the median time between funding rounds has lengthened, more startups have shut down through bankruptcy, and the share of "down rounds" — new financing at a lower valuation — rose from 8% to nearly 20%, with those rounds cutting post-money valuations by more than 50%.
  • AI Investments Are Booming, but Venture-Firm Profits Are at a Historic Low — U.S. venture firms returned $26bn worth of shares to their investors in a year, the lowest amount since 2011, according to data provider PitchBook, and invested $60bn more than they collected — the largest such deficit in 26 years of PitchBook data. Investment remains high while acquisitions and initial public offerings are few.

AI Summary. Averaging corporate earnings over 10 years to strip out the business cycle shows US stocks are expensive, with the cyclically adjusted price-to-earnings ratio at 38, near all-time highs and well above its long-term average of 17.

Torsten Sløk Apollo
Date Posted:

The S&P 500 Shiller CAPE stands at 38, significantly above its long term mean of 17 and near the all-time high of 44. Torsten Sløk cautions “stocks are expensive.”

Are US stocks expensive after adjusting earnings for the business cycle?

Core argument: The cyclically adjusted price-earnings ratio stands at 38, near all-time highs and well above its long-term average of 17, which indicates stocks are very expensive.

When growth is strong, corporate earnings are high. When growth is weak, corporate earnings are low. This makes it difficult to find out if companies are cheap or expensive. One way to analyze if stocks are cheap or expensive is to remove the business cycle by taking the 10-year average of earnings, and doing so shows that stocks are very expensive at the moment. Specifically, the cyclically adjusted price earnings ratio at 38 is near all-time highs, significantly above its long-term average at 17.

Takeaways by Macro Roundup® AI

  1. The cyclically adjusted price-earnings ratio stands at 38, near all-time highs and well above its long-term average of 17, which indicates stocks are very expensive.
  2. Because corporate earnings rise when growth is strong and fall when growth is weak, it is difficult to judge whether companies are cheap or expensive.
  3. Taking a 10-year average of earnings removes the business cycle from valuation analysis, and that adjustment shows stocks are very expensive.

Related Articles:

AI Summary. Researchers developed a recurrent, transformer-based neural network that decodes the surface code, the leading quantum error-correction code, and report it outperforms other state-of-the-art decoders on data from Google's Sycamore quantum processor and on simulated data with realistic noise. The authors argue machine learning can surpass human-designed decoding algorithms by learning directly from data.

Johannes Bausch, Andrew Senior, Francisco Heras et al. Nature
Date Posted:

Google DeepMind announces AlphaQubit, an AI-based decoder that identifies quantum computing errors with state-of-the art accuracy, addressing one of the biggest challenges in scaling quantum computing.

Can machine learning decode quantum errors better than human-designed algorithms?

Core argument: A recurrent, transformer-based neural network decoder outperforms other state-of-the-art decoders on real-world data from Google's Sycamore quantum processor for distance 3 and distance 5 surface codes, the leading quantum error-correction code.

Building a large-scale quantum computer requires effective strategies to correct errors that inevitably arise in physical quantum systems. Quantum error-correction codes present a way to reach this goal by encoding logical information redundantly into many physical qubits. Here we develop a recurrent, transformer-based neural network that learns to decode the surface code, the leading quantum error-correction code. Our decoder outperforms other state-of-the-art decoders on real-world data from Google’s Sycamore quantum processor for distance 3 and distance 5 surface codes. On distances up to 11, the decoder maintains its advantage on simulated data with realistic noise including cross-talk and leakage, utilizing soft readouts and leakage information. Our work illustrates the ability of machine learning to go beyond human-designed algorithms by learning from data directly, highlighting machine learning as a strong contender for decoding in quantum computers.

Takeaways by Macro Roundup® AI

  1. A recurrent, transformer-based neural network decoder outperforms other state-of-the-art decoders on real-world data from Google's Sycamore quantum processor for distance 3 and distance 5 surface codes, the leading quantum error-correction code.
  2. On simulated data with realistic noise including cross-talk and leakage, the neural network decoder maintains its advantage at code distances up to 11 by using soft readouts and leakage information.
  3. The authors argue that machine learning can go beyond human-designed algorithms by learning from data directly, making it a strong contender for decoding in quantum computers.

Related Articles:

  • Quantum Computing Advance Begins New Era, IBM Says — IBM researchers report that a 127-qubit quantum processor simulated the magnetic behavior of 127 atom-scale bar magnets, a problem too complex for an exact answer on the largest supercomputers. Because quantum noise made each sub-millisecond calculation unreliable, the computation was repeated 600,000 times to converge on an answer.

AI Summary. U.S. intelligence and national security officials have told major telecommunications companies to find a permanent way to expel Chinese hackers from the deepest parts of the nation's communications networks. Critical parts of the system date to the late 1970s or early 1980s and are too old to upgrade with modern cybersecurity protections.

David Sanger, Julian Barnes, Devlin Barrett and Adam Goldman New York Times
Date Posted:

MSS’s “Salt Typhoon” exploit may require rebuilding the basic infrastructure of the American telecommunications system in order to evict Chinese intelligence from the system.

Is aging telecommunications infrastructure making Chinese hackers hard to remove?

Core argument: Top U.S. intelligence and national security officials summoned leaders of major telecommunications companies to the White House to press them to help find a permanent way to expel Chinese hackers from the nation's communications networks.

Leaders of the top telecommunications companies were summoned to the White House on Friday to discuss a security problem that has been roiling the government: how to expel Chinese hackers from the deepest corners of the nation’s communications networks. At the White House meeting, the message delivered by top American intelligence and national security officials was that despite the aging technology, the telecommunications companies needed to help find a permanent way to keep Beijing’s agents out of the systems. Critical parts of the American telecommunications system are too old to upgrade with modern cybersecurity protections. Some parts of the system date to the late 1970s or early 1980s, when landlines, not cellphones, dominated the network. One participant in the meeting said the only solution to the problem was “ripping out and replacing whole sections of the networks,” a process the companies have been slow to invest in.

Takeaways by Macro Roundup® AI

  1. Top U.S. intelligence and national security officials summoned leaders of major telecommunications companies to the White House to press them to help find a permanent way to expel Chinese hackers from the nation's communications networks.
  2. Critical parts of the American telecommunications system date to the late 1970s or early 1980s, when landlines dominated the network, and are too old to upgrade with modern cybersecurity protections.
  3. One participant in the White House meeting said the only solution was "ripping out and replacing whole sections of the networks," an investment the telecommunications companies have been slow to make.

Related Articles:

  • U.S. Officials Race to Understand Severity of China’s Salt Typhoon Hacks — Federal authorities are investigating breaches of Verizon, AT&T and Lumen networks attributed to Salt Typhoon, a hacking group tied to Chinese intelligence. Investigators believe the compromises may have given the hackers access to systems the U.S. government uses for court-authorized wiretapping requests, though the severity remains unclear.
  • Volt Typhoon Targets US Critical Infrastructure with Living-Off-The-Land Techniques — Microsoft has identified stealthy intrusions by Volt Typhoon, a Chinese state-sponsored actor, against US critical infrastructure organizations spanning communications, manufacturing, utilities, transportation, maritime, government, information technology, and education. Microsoft assesses with moderate confidence that the campaign is developing capabilities that could disrupt critical communications infrastructure between the US and Asia during future crises.
  • Xi Jinping Says He Is Preparing China for War — China's leadership has emphasized war readiness, announcing a 7.2% increase in the defense budget—which has doubled over the past decade—and plans to reduce dependence on foreign grain imports. China has also opened National Defense Mobilization recruitment offices across the country, while cities in Fujian Province opposite Taiwan build or upgrade air-raid shelters.