REUTERS
"Imports hold back US economy in first quarter, inflation flares up"
By Lucia Mutikani
April 25, 2024
Summary
* First-quarter GDP increases at 1.6% rate
* Consumer spending slows, trade deficit widens
* Core PCE inflation accelerates at 3.7% pace
* Weekly jobless claims fall 5,000 to 207,000
WASHINGTON, April 25 (Reuters) - The U.S. economy grew at its slowest pace in nearly two years in the first quarter amid a surge in imports and small build-up of unsold goods at businesses, signs of solid demand that together with an acceleration in inflation reinforced expectations the Federal Reserve would not cut interest rates before September.
The cooler-than-expected growth reported by the Commerce Department in its snapshot of first-quarter gross domestic product on Thursday, which also reflected a downshift in government spending, exaggerated the moderation in economic activity.
Domestic demand, a better growth measure, was strong as consumer spending moderated slightly while business investment picked up and the housing recovery gained steam.
Trade and inventories are the most volatile GDP components, and are often subject to revision when the government updates its growth estimates.
Fed officials are expected to leave rates unchanged at the U.S. central bank's policy meeting next week.
"The Fed will likely see the GDP report as solid, while the upward surprise to inflation will support the central bank's case for waiting longer before cutting rates," said Daniel Vernazza, chief international economist at UniCredit.
GDP increased at a 1.6% annualized rate last quarter, the slowest pace since the second quarter of 2022, the Commerce Department's Bureau of Economic Analysis said.
Economists polled by Reuters had forecast GDP would rise at a 2.4% rate, with estimates ranging from a 1.0% pace to a 3.1% rate.
The economy grew at a 3.4% rate in the fourth quarter.
The first-quarter growth pace was below what U.S. central bank officials regard as the non-inflationary growth rate of 1.8%.
Excluding inventories, government spending and trade, the economy grew at a 3.1% rate after expanding at a 3.3% rate in the fourth quarter.
That also dispels the notion that government spending was fueling the economy.
The U.S. economy, which has outperformed the economies of other advanced nations, is being supported by a resilient labor market.
U.S. Treasury Secretary Janet Yellen told Reuters in an interview that she was focused on consumer and business spending.
"Those two elements of final demand came in line with last year's growth rate ... so this is the underlying strength of the U.S. economy that showed continuing robust strength and an economy firing on all cylinders."
Price pressures heated up by the most in a year, with a measure of inflation in the economy increasing at a 3.1% rate after rising at a 1.9% pace in the October-December quarter.
The personal consumption expenditures (PCE) price index excluding food and energy surged at a 3.7% rate after increasing at a 2.0% pace in the fourth quarter.
The so-called core PCE price index is one of the inflation measures tracked by the Fed for its 2% target.
Inflation was boosted by increases in the costs of services like transportation, insurance and housing, which offset a decline in goods prices such as motor vehicles and parts.
The strong readings pose an upside risk to March PCE inflation data due to be released on Friday, though much would depend on revisions to the January and February data.
The Fed has kept its benchmark overnight interest rate in the 5.25%-5.50% range since July.
It has raised the policy rate by 525 basis points since March of 2022.
Stocks on Wall Street were trading lower.
The dollar slipped against a basket of currencies.
U.S. Treasury yields rose.
TIGHT LABOR MARKET
A significant slowdown in the labor market is not yet evident.
The Labor Department's weekly jobless claims report showed initial claims for unemployment benefits fell 5,000 to a seasonally adjusted 207,000 in the week ending April 20.
The number of people receiving benefits after an initial week of aid, a proxy for hiring, declined 15,000 to 1.781 million during the week ending April 13.
The so-called continuing claims data covered the period during which the government surveyed households for April's unemployment rate.
Continuing claims fell between the March and April survey periods, implying the unemployment rate was likely unchanged after dipping to 3.8% last month from 3.9% in February.
Low layoffs are keeping wages high, sustaining consumer spending, which accounts for more than two-thirds of economic activity.
Consumer spending grew at a still-solid 2.5% rate, slowing from the 3.3% growth pace rate notched in the October-December quarter.
Spending was driven by healthcare, financial services and insurance, which more than offset a decline in goods, including motor vehicles and gasoline.
Spending is likely to gradually cool this year.
Lower-income households have depleted their COVID-19 pandemic savings and are largely relying on debt to fund purchases.
Recent data and comments from bank executives indicated that lower-income borrowers were increasingly struggling to keep up with their loan payments.
Though income increased at a $407.1 billion rate compared with the fourth quarter's $230.2 billion pace, the gains were eroded by inflation and higher taxes.
Income at the disposal of households after accounting for inflation and taxes rose at a 1.1% rate versus a 2.0% pace in the October-December quarter.
The saving rate decreased to 3.6% from 4.0% in the prior quarter.
"The recent stickiness in inflation lends downside risk to the near-term forecast for consumption as it could weigh on real disposable income," said Ryan Sweet, chief economist at Oxford Economics.
Inventories were whittled down, rising at a $35.4 billion rate after increasing at a $54.9 billion pace in the fourth quarter.
Inventories subtracted 0.35 percentage point from GDP growth.
Part of the spending was satiated with imports, which resulted in the trade deficit widening to $973.2 billion from $918.5 billion in the October-December quarter.
Trade chopped off 0.86 percentage point from GDP growth.
Government spending decelerated to a 1.2% rate from the 4.6% pace notched in the October-December quarter amid a decline in federal government outlays, mostly defense.
Business spending picked up as companies invested in artificial intelligence.
Investment in nonresidential structures like factories contracted for the first time in more than year as the boost from policies by the Biden administration to bring the production of semiconductor manufacturing back to the U.S. faded.
Residential investment recorded its fastest pace of growth since the fourth quarter of 2020, thanks to rising home sales and housing construction, despite higher mortgage rates.
"Don't underestimate this economy," said Shannon Grein, an economist at Well Fargo.
Reporting by Lucia Mutikani; additional reporting by Alessandra Galloni; Editing by Chizu Nomiyama, Andrea Ricci and Paul Simao
https://www.reuters.com/markets/us/us-e ... 024-04-25/
POLITICS
Re: POLITICS
REUTERS
"Yellen says US economy strong, all options open on China's overcapacity"
By Alessandra Galloni, David Lawder and Andrea Shalal
April 25, 2024
Summary
* Yellen: Q1 GDP data may be revised up; spending, investment strong
* Inflation to continue to ebb despite Q1 flare-up, Yellen says
* Yellen: tapping interest on Russian assets can win G7 support
WASHINGTON, April 25 (Reuters) - U.S. Treasury Secretary Janet Yellen told Reuters on Thursday that U.S. economic growth was likely stronger than suggested by weaker-than-expected data on first-quarter output and said the Biden administration was keeping all options open to respond to threats from China's excess industrial capacity.
In a wide-ranging Reuters Next interview, Yellen also said that a U.S. proposal for using the interest earnings from $300 billion in frozen Russian assets to aid Ukraine could win broad support from G7 allies.
Yellen said U.S. GDP growth for the first quarter could be revised higher after more data is in hand and inflation will ease to more normal levels after a clutch of "peculiar" factors held the economy to its weakest showing in nearly two years.
"The U.S. economy continues to perform very, very well," Yellen said in an interview with Reuters, responding to the Commerce Department's report showing that U.S. gross domestic product grew at a 1.6% annualized rate last quarter.
That was below the 2.4% estimated by economists and less than half the pace in the fourth quarter of 2023 - thanks to substantial drags from trade and private inventories.
The report also showed a worrisome surge in inflation, with the personal consumption expenditures (PCE) price index excluding food and energy rising at a 3.7% annual rate after a 2.0% pace in the fourth quarter of 2023.
Yellen downplayed the inflation jump and said she did not see that as indicating that unemployment needed to increase or other areas of the economy needed to cool to return inflation to the Fed's 2% target.
"The fundamentals here are in line with inflation continuing back down to normal levels," Yellen said.
Fighting inflation remained President Joe Biden's top priority, Yellen said, highlighting his administration's efforts to reduce healthcare, energy and housing costs.
But Biden, a Democrat, has struggled to translate U.S. economic strength into voter support ahead of the November presidential election.
Republican challenger Donald Trump led Biden by seven percentage points in a recent Reuters/Ipsos poll when voters were asked which candidate would be better for the economy.
"What I focus on most is the strength of consumer spending and investment spending," Yellen said.
"Those two elements of final demand came in in line with last year's growth rate ... so this is the underlying strength of the U.S. economy that showed continuing robust strength."
"The headline figure was off a little bit but for reasons that are peculiar and not really indicative of underlying strength," she added.
Indeed, a number of private economists said the GDP data likely overstated any weakness in an economy that had grown at above the rate most see as its potential for nearly two years, despite aggressive interest rate hikes over that span by the U.S. Federal Reserve aimed at quashing inflation.
Yellen said dollar strength has been another byproduct of U.S. growth and tight monetary policy.
She acknowledged that this has put some pressure on other countries, but said currency interventions should occur only in "very rare and exceptional circumstances," when markets are disorderly with excessive volatility.
She declined to comment on the Japanese yen's value when asked whether it was out of line with fundamentals.
Last week, the U.S., Japan and South Korea agreed to consult closely on currencies, acknowledging concerns from Tokyo and Seoul over their currencies recent sharp declines against the dollar.
CHINA OVERCAPACITY
Yellen told Reuters no option was "off the table" for dealing with one threat to the U.S. economy - overproduction in China, which was hurting manufacturers in numerous countries.
She said that while Chinese policymakers have acknowledged they have a problem with excess industrial capacity for electric vehicles, solar panels and other clean energy goods, they need to address it.
The issue was "discussed intensively" last week at a U.S.-China meeting on the sidelines of the International Monetary Fund and World Bank spring meetings in Washington, she said.
Asked about potential for new tariffs or other actions to protect U.S. producers from an expected flood of Chinese exports, Yellen said she would not eliminate any options as a possible response.
She said Chinese overproduction threatens the viability of manufacturers in the U.S., Europe, Japan, Mexico and India but the problem won't be resolved "in a day or a week."
"So it's important that China recognize the concern and begin to act to address it," Yellen said.
"But we don't want our industry wiped out in the meantime, so I wouldn't want to take anything off the table."
The Biden administration is completing a review of the "Section 301" unfair trade tariffs on Chinese imports imposed by former President Donald Trump in 2018, which U.S. officials have said could lead to higher tariffs on some products.
Biden last week called for the review to triple the Section 301 duties on Chinese steel to 25%.
U.S. Trade Representative Katherine Tai also told senators that the U.S. needed to take "early action, decisive action" to protect the fledgling American EV sector from Chinese imports.
U.S. tariffs on Chinese vehicle imports are now about 27.5%, and few Chinese EVs are sold in the U.S. at the moment.
RUSSIAN ASSET PLANS
Yellen said that a proposal under discussion by finance ministers from the Group of Seven (G7) industrial democracies to harness earnings from frozen Russian central bank assets to aid Ukraine can be achieved without an outright confiscation of those assets, allaying the concerns of some countries.
Yellen welcomed what she called a "very constructive step" taken by the European Union to segregate the proceeds from assets held by Brussels-based Euroclear and transfer them to Ukraine, noting future interest could also be pulled forward to back loans to Ukraine.
"This is an approach that could be broadly supported by countries that are concerned about the seizure of assets, and some of the interest could be brought forward through, for example, a loan," Yellen said.
Yellen said the approach was among several options being discussed by G7 countries ahead of a leaders summit in June, adding, "it certainly belongs on the list."
The U.S. approach, led by deputy national security adviser Daleep Singh, is gaining momentum among the G7 nations, two officials from the group told Reuters earlier on Thursday.
Most of the Russian assets held by Euroclear have now been converted to cash, Yellen told Reuters. G7 officials say the assets could generate around $5 billion a year in interest.
Reporting by Alessandra Galloni, David Lawder and Andrea Shalal; Additional reporting by Dan Burns and Lindsay Dunsmuir; Writing by David Lawder, Andrea Shalal and Dan Burns; Editing by Andrea Ricci
https://www.reuters.com/markets/us/yell ... 024-04-25/
"Yellen says US economy strong, all options open on China's overcapacity"
By Alessandra Galloni, David Lawder and Andrea Shalal
April 25, 2024
Summary
* Yellen: Q1 GDP data may be revised up; spending, investment strong
* Inflation to continue to ebb despite Q1 flare-up, Yellen says
* Yellen: tapping interest on Russian assets can win G7 support
WASHINGTON, April 25 (Reuters) - U.S. Treasury Secretary Janet Yellen told Reuters on Thursday that U.S. economic growth was likely stronger than suggested by weaker-than-expected data on first-quarter output and said the Biden administration was keeping all options open to respond to threats from China's excess industrial capacity.
In a wide-ranging Reuters Next interview, Yellen also said that a U.S. proposal for using the interest earnings from $300 billion in frozen Russian assets to aid Ukraine could win broad support from G7 allies.
Yellen said U.S. GDP growth for the first quarter could be revised higher after more data is in hand and inflation will ease to more normal levels after a clutch of "peculiar" factors held the economy to its weakest showing in nearly two years.
"The U.S. economy continues to perform very, very well," Yellen said in an interview with Reuters, responding to the Commerce Department's report showing that U.S. gross domestic product grew at a 1.6% annualized rate last quarter.
That was below the 2.4% estimated by economists and less than half the pace in the fourth quarter of 2023 - thanks to substantial drags from trade and private inventories.
The report also showed a worrisome surge in inflation, with the personal consumption expenditures (PCE) price index excluding food and energy rising at a 3.7% annual rate after a 2.0% pace in the fourth quarter of 2023.
Yellen downplayed the inflation jump and said she did not see that as indicating that unemployment needed to increase or other areas of the economy needed to cool to return inflation to the Fed's 2% target.
"The fundamentals here are in line with inflation continuing back down to normal levels," Yellen said.
Fighting inflation remained President Joe Biden's top priority, Yellen said, highlighting his administration's efforts to reduce healthcare, energy and housing costs.
But Biden, a Democrat, has struggled to translate U.S. economic strength into voter support ahead of the November presidential election.
Republican challenger Donald Trump led Biden by seven percentage points in a recent Reuters/Ipsos poll when voters were asked which candidate would be better for the economy.
"What I focus on most is the strength of consumer spending and investment spending," Yellen said.
"Those two elements of final demand came in in line with last year's growth rate ... so this is the underlying strength of the U.S. economy that showed continuing robust strength."
"The headline figure was off a little bit but for reasons that are peculiar and not really indicative of underlying strength," she added.
Indeed, a number of private economists said the GDP data likely overstated any weakness in an economy that had grown at above the rate most see as its potential for nearly two years, despite aggressive interest rate hikes over that span by the U.S. Federal Reserve aimed at quashing inflation.
Yellen said dollar strength has been another byproduct of U.S. growth and tight monetary policy.
She acknowledged that this has put some pressure on other countries, but said currency interventions should occur only in "very rare and exceptional circumstances," when markets are disorderly with excessive volatility.
She declined to comment on the Japanese yen's value when asked whether it was out of line with fundamentals.
Last week, the U.S., Japan and South Korea agreed to consult closely on currencies, acknowledging concerns from Tokyo and Seoul over their currencies recent sharp declines against the dollar.
CHINA OVERCAPACITY
Yellen told Reuters no option was "off the table" for dealing with one threat to the U.S. economy - overproduction in China, which was hurting manufacturers in numerous countries.
She said that while Chinese policymakers have acknowledged they have a problem with excess industrial capacity for electric vehicles, solar panels and other clean energy goods, they need to address it.
The issue was "discussed intensively" last week at a U.S.-China meeting on the sidelines of the International Monetary Fund and World Bank spring meetings in Washington, she said.
Asked about potential for new tariffs or other actions to protect U.S. producers from an expected flood of Chinese exports, Yellen said she would not eliminate any options as a possible response.
She said Chinese overproduction threatens the viability of manufacturers in the U.S., Europe, Japan, Mexico and India but the problem won't be resolved "in a day or a week."
"So it's important that China recognize the concern and begin to act to address it," Yellen said.
"But we don't want our industry wiped out in the meantime, so I wouldn't want to take anything off the table."
The Biden administration is completing a review of the "Section 301" unfair trade tariffs on Chinese imports imposed by former President Donald Trump in 2018, which U.S. officials have said could lead to higher tariffs on some products.
Biden last week called for the review to triple the Section 301 duties on Chinese steel to 25%.
U.S. Trade Representative Katherine Tai also told senators that the U.S. needed to take "early action, decisive action" to protect the fledgling American EV sector from Chinese imports.
U.S. tariffs on Chinese vehicle imports are now about 27.5%, and few Chinese EVs are sold in the U.S. at the moment.
RUSSIAN ASSET PLANS
Yellen said that a proposal under discussion by finance ministers from the Group of Seven (G7) industrial democracies to harness earnings from frozen Russian central bank assets to aid Ukraine can be achieved without an outright confiscation of those assets, allaying the concerns of some countries.
Yellen welcomed what she called a "very constructive step" taken by the European Union to segregate the proceeds from assets held by Brussels-based Euroclear and transfer them to Ukraine, noting future interest could also be pulled forward to back loans to Ukraine.
"This is an approach that could be broadly supported by countries that are concerned about the seizure of assets, and some of the interest could be brought forward through, for example, a loan," Yellen said.
Yellen said the approach was among several options being discussed by G7 countries ahead of a leaders summit in June, adding, "it certainly belongs on the list."
The U.S. approach, led by deputy national security adviser Daleep Singh, is gaining momentum among the G7 nations, two officials from the group told Reuters earlier on Thursday.
Most of the Russian assets held by Euroclear have now been converted to cash, Yellen told Reuters. G7 officials say the assets could generate around $5 billion a year in interest.
Reporting by Alessandra Galloni, David Lawder and Andrea Shalal; Additional reporting by Dan Burns and Lindsay Dunsmuir; Writing by David Lawder, Andrea Shalal and Dan Burns; Editing by Andrea Ricci
https://www.reuters.com/markets/us/yell ... 024-04-25/
Re: POLITICS
REUTERS
"Intel forecasts second-quarter revenue below estimates, shares fall"
By Arsheeya Bajwa and Max A. Cherney
April 25, 2024
April 25 (Reuters) - Intel forecast second-quarter revenue and profit below market estimates on Thursday, as it faces weak demand for its traditional data center and personal computer chips amid efforts to build its contract manufacturing business.
Shares of the Santa Clara, California-based company fell close to 6% in extended trading.
The company expects second-quarter revenue in the range of $12.5 billion to $13.5 billion, compared with analysts' average estimate of $13.57 billion, according to LSEG data.
Intel forecast second-quarter adjusted earnings of 10 cents per share, also below expectations.
Enterprises have prioritized spending on advanced and speedy AI server chips, hurting demand for Intel's central processing units (CPUs), which have been the mainstay chip powering data centers for decades.
According to analysts, surging demand for graphics processing units (GPUs) useful for AI applications has also reduced the appetite for CPUs, which are Intel's main product.
Intel's total revenue of $12.72 billion in the first quarter marginally missed expectations of $12.78 billion.
Sales at its data center business rose 5% to $3 billion during the period.
Nvidia's powerful GPUs dominate the AI market, as large and small companies have sought to acquire tens of billions of dollars worth of its processors.
Still, constrained supply for these advanced and speedy processors has left Intel with the opportunity to capitalize on the towering demand, though it is a late entrant in the market.
Intel forecasts more than $500 million in revenue from its Gaudi AI chips this year, CEO Patrick Gelsinger told Reuters on Thursday.
Preliminary results from IDC showed the PC market returned to growth in the first quarter after about two years of declines.
Revenue from the client computing segment, which houses Intel's PC chips, rose 31% during the quarter.
Intel's contract manufacturing business, or foundry, is working to catch industry leader TSMC, but its profitability is years away.
Revenue from the foundry business fell 10% in the first quarter.
The company recorded an adjusted gross margin of 45.1% for the first quarter, compared with analysts' expectation of 44.3%.
Inter-segment eliminations of $4.4 billion were made to prevent double-counting of revenue.
Reporting by Arsheeya Bajwa in Bengaluru; Editing by Shilpi Majumdar
https://www.reuters.com/technology/inte ... 024-04-25/
"Intel forecasts second-quarter revenue below estimates, shares fall"
By Arsheeya Bajwa and Max A. Cherney
April 25, 2024
April 25 (Reuters) - Intel forecast second-quarter revenue and profit below market estimates on Thursday, as it faces weak demand for its traditional data center and personal computer chips amid efforts to build its contract manufacturing business.
Shares of the Santa Clara, California-based company fell close to 6% in extended trading.
The company expects second-quarter revenue in the range of $12.5 billion to $13.5 billion, compared with analysts' average estimate of $13.57 billion, according to LSEG data.
Intel forecast second-quarter adjusted earnings of 10 cents per share, also below expectations.
Enterprises have prioritized spending on advanced and speedy AI server chips, hurting demand for Intel's central processing units (CPUs), which have been the mainstay chip powering data centers for decades.
According to analysts, surging demand for graphics processing units (GPUs) useful for AI applications has also reduced the appetite for CPUs, which are Intel's main product.
Intel's total revenue of $12.72 billion in the first quarter marginally missed expectations of $12.78 billion.
Sales at its data center business rose 5% to $3 billion during the period.
Nvidia's powerful GPUs dominate the AI market, as large and small companies have sought to acquire tens of billions of dollars worth of its processors.
Still, constrained supply for these advanced and speedy processors has left Intel with the opportunity to capitalize on the towering demand, though it is a late entrant in the market.
Intel forecasts more than $500 million in revenue from its Gaudi AI chips this year, CEO Patrick Gelsinger told Reuters on Thursday.
Preliminary results from IDC showed the PC market returned to growth in the first quarter after about two years of declines.
Revenue from the client computing segment, which houses Intel's PC chips, rose 31% during the quarter.
Intel's contract manufacturing business, or foundry, is working to catch industry leader TSMC, but its profitability is years away.
Revenue from the foundry business fell 10% in the first quarter.
The company recorded an adjusted gross margin of 45.1% for the first quarter, compared with analysts' expectation of 44.3%.
Inter-segment eliminations of $4.4 billion were made to prevent double-counting of revenue.
Reporting by Arsheeya Bajwa in Bengaluru; Editing by Shilpi Majumdar
https://www.reuters.com/technology/inte ... 024-04-25/
Re: POLITICS
REUTERS
"Biden announces preliminary deal with Micron for up to $6.14 bln in chip grants"
By Jarrett Renshaw and Nandita Bose
April 25, 2024
WASHINGTON, April 25 (Reuters) - U.S. President Joe Biden traveled to Syracuse, New York, on Thursday and announced a preliminary agreement with memory chip maker Micron Technology for up to $6.14 billion in subsidies for two chip factories.
The agreement signed by the U.S. Commerce Department will fund facilities in New York and Idaho under the 2022 CHIPS and Science Act, which aims to boost domestic manufacturing of chips and reduce reliance on supplies from China and Taiwan.
Biden said the United States used to have 40% of the chips market but over time production moved outside the country and the pandemic exposed weaknesses in the U.S. supply chain that hurt critical industries.
"I'm determined that I'm never going to let us be vulnerable to wait lines again, what is essential is we're going to make it here in America together," Biden said.
The Commerce Department said the federal grants would support the construction of a fabrication plant, or fab, in Clay, New York, a first step toward Micron's plans to invest about $100 billion in New York and create 13,500 jobs.
The grants also provide initial funding for a facility in Boise, Idaho, unlocking a planned $25 billion investment in a fab to be co-located with Micron's research and development facilities there and should create 6,500 jobs, Commerce said.
Micron's investment will be the "largest private investment in New York and Idaho's history," and will create over 70,000 jobs, including 20,000 direct construction and manufacturing jobs and tens of thousands of indirect jobs," the White House said.
Biden, who is running for re-election in November's presidential election, used his visit to Syracuse to tout his administration's efforts to revitalize U.S. manufacturing and strengthen national security.
"American manufacturing is back, new factories are going up all across the country, and communities like Syracuse are writing the great American comeback story," Biden said.
In the evening, the president will speak at a campaign event in Westchester County, New York.
Biden signed the $52.7 billion CHIPS bill in August 2022 to subsidize U.S. semiconductor production and research.
Semiconductors were invented in the United States, but domestic companies produce only about 10% of the world's chips and none of the most advanced ones.
The White House said Thursday's announcement also included at least $40 million in funds for training and workforce development, as well as creation of four more workforce hubs in upstate New York, Milwaukee, Philadelphia and Michigan.
Under the agreement, Micron committed to providing affordable high-quality childcare for its workers across its facilities.
The company also affirmed "workers' rights to organize, to share feedback without fear of reprisal, and to collectively bargain," the White House said.
Reporting by Jarrret Renshaw, Nandita Bose and Andrea Shalal; Editing by Tom Hogue and Diane Craft
https://www.reuters.com/technology/bide ... 024-04-25/
"Biden announces preliminary deal with Micron for up to $6.14 bln in chip grants"
By Jarrett Renshaw and Nandita Bose
April 25, 2024
WASHINGTON, April 25 (Reuters) - U.S. President Joe Biden traveled to Syracuse, New York, on Thursday and announced a preliminary agreement with memory chip maker Micron Technology for up to $6.14 billion in subsidies for two chip factories.
The agreement signed by the U.S. Commerce Department will fund facilities in New York and Idaho under the 2022 CHIPS and Science Act, which aims to boost domestic manufacturing of chips and reduce reliance on supplies from China and Taiwan.
Biden said the United States used to have 40% of the chips market but over time production moved outside the country and the pandemic exposed weaknesses in the U.S. supply chain that hurt critical industries.
"I'm determined that I'm never going to let us be vulnerable to wait lines again, what is essential is we're going to make it here in America together," Biden said.
The Commerce Department said the federal grants would support the construction of a fabrication plant, or fab, in Clay, New York, a first step toward Micron's plans to invest about $100 billion in New York and create 13,500 jobs.
The grants also provide initial funding for a facility in Boise, Idaho, unlocking a planned $25 billion investment in a fab to be co-located with Micron's research and development facilities there and should create 6,500 jobs, Commerce said.
Micron's investment will be the "largest private investment in New York and Idaho's history," and will create over 70,000 jobs, including 20,000 direct construction and manufacturing jobs and tens of thousands of indirect jobs," the White House said.
Biden, who is running for re-election in November's presidential election, used his visit to Syracuse to tout his administration's efforts to revitalize U.S. manufacturing and strengthen national security.
"American manufacturing is back, new factories are going up all across the country, and communities like Syracuse are writing the great American comeback story," Biden said.
In the evening, the president will speak at a campaign event in Westchester County, New York.
Biden signed the $52.7 billion CHIPS bill in August 2022 to subsidize U.S. semiconductor production and research.
Semiconductors were invented in the United States, but domestic companies produce only about 10% of the world's chips and none of the most advanced ones.
The White House said Thursday's announcement also included at least $40 million in funds for training and workforce development, as well as creation of four more workforce hubs in upstate New York, Milwaukee, Philadelphia and Michigan.
Under the agreement, Micron committed to providing affordable high-quality childcare for its workers across its facilities.
The company also affirmed "workers' rights to organize, to share feedback without fear of reprisal, and to collectively bargain," the White House said.
Reporting by Jarrret Renshaw, Nandita Bose and Andrea Shalal; Editing by Tom Hogue and Diane Craft
https://www.reuters.com/technology/bide ... 024-04-25/
Re: POLITICS
"Biden in Syracuse to celebrate Micron's $6.1 billion CHIPS grant - President's visit comes after two law enforcement officers were slain in a shootout in Salina, eight miles from Micron's campus"
By Larry Rulison, Albany, New York Times Union
April 25, 2024
ALBANY – President Joe Biden landed in Syracuse on Thursday to celebrate a $6.1 billion grant awarded to Micron Technology, the memory chipmaker that is building a $100 billion manufacturing campus in the town of Clay, just north of the city.
Biden was joined by Gov. Kathy Hochul and U.S. Senate Majority Leader Charles Schumer of New York at the Milton J. Rubenstein Museum of Science & Technology in Syracuse's Armory Square.
In addition to touting the Micron award, being awarded through the $52 billion CHIPS and Science Act, the president was planning to meet with the families of the two police officers killed in a shootout on April 14 in the town of Salina, about six miles north of where Biden was speaking.
Funerals for the two men, Lt. Michael Hoosock of the Onondaga County Sheriff's Department, and Syracuse police Officer Michael Jensen, were held last week.
Both departments and their members were still reeling from the tragedy.
As a result, State Troopers were tasked with providing security for the president's visit instead of local law enforcement, which criticized the White House for not rescheduling the trip.
The $6.1 billion grant Micron will receive is coming from the $52 billion CHIPS and Science Act, the chip industry subsidy program that was authored by Schumer and designed to revitalize the domestic chip sector in the face of China seeking to dominate the industry and its supply chain.
Micron's award, which also includes $7 billion in federal loans, is the third CHIPS grant to be announced for large manufacturers.
The Micron event in Syracuse comes a month after Schumer and Hochul celebrated a $1.5 billion CHIPS grant that was awarded to GlobalFoundries to assist the company's construction of a second factory at the site, which supplies chips to the military, automakers and telecommunications companies among other industries.
The grant would also go toward an expansion at GlobalFoundries' chip factory in Essex Junction, Vt. outside Burlington.
The $6.1 billion grant for Micron will be in addition to $7.5 billion in loans that the company can access.
The funding is not only for the Syracuse-area project but also a new fab, or factory, that Micron is planning in its home state of Idaho.
https://www.timesunion.com/business/art ... 0headlines
By Larry Rulison, Albany, New York Times Union
April 25, 2024
ALBANY – President Joe Biden landed in Syracuse on Thursday to celebrate a $6.1 billion grant awarded to Micron Technology, the memory chipmaker that is building a $100 billion manufacturing campus in the town of Clay, just north of the city.
Biden was joined by Gov. Kathy Hochul and U.S. Senate Majority Leader Charles Schumer of New York at the Milton J. Rubenstein Museum of Science & Technology in Syracuse's Armory Square.
In addition to touting the Micron award, being awarded through the $52 billion CHIPS and Science Act, the president was planning to meet with the families of the two police officers killed in a shootout on April 14 in the town of Salina, about six miles north of where Biden was speaking.
Funerals for the two men, Lt. Michael Hoosock of the Onondaga County Sheriff's Department, and Syracuse police Officer Michael Jensen, were held last week.
Both departments and their members were still reeling from the tragedy.
As a result, State Troopers were tasked with providing security for the president's visit instead of local law enforcement, which criticized the White House for not rescheduling the trip.
The $6.1 billion grant Micron will receive is coming from the $52 billion CHIPS and Science Act, the chip industry subsidy program that was authored by Schumer and designed to revitalize the domestic chip sector in the face of China seeking to dominate the industry and its supply chain.
Micron's award, which also includes $7 billion in federal loans, is the third CHIPS grant to be announced for large manufacturers.
The Micron event in Syracuse comes a month after Schumer and Hochul celebrated a $1.5 billion CHIPS grant that was awarded to GlobalFoundries to assist the company's construction of a second factory at the site, which supplies chips to the military, automakers and telecommunications companies among other industries.
The grant would also go toward an expansion at GlobalFoundries' chip factory in Essex Junction, Vt. outside Burlington.
The $6.1 billion grant for Micron will be in addition to $7.5 billion in loans that the company can access.
The funding is not only for the Syracuse-area project but also a new fab, or factory, that Micron is planning in its home state of Idaho.
https://www.timesunion.com/business/art ... 0headlines
Re: POLITICS
CNBC
"Key Fed inflation measure rose 2.8% in March from a year ago, more than expected"
Jeff Cox @JEFF.COX.7528 @JEFFCOXCNBCCOM
PUBLISHED FRI, APR 26 2024
KEY POINTS
* The core personal consumption expenditures price index excluding food and energy increased 2.8% from a year ago in March, unchanged from February and slightly higher than expected.
* Personal spending rose 0.8% on the month, more than the personal income increase of 0.5%.
* The personal saving rate fell to 3.2%, down 0.4 percentage points from February and 2 full percentage points from a year ago.
Inflation showed few signs of letting up in March, with a key barometer the Federal Reserve watches closely showing that price pressures remain elevated.
The personal consumption expenditures price index excluding food and energy increased 2.8% from a year ago in March, the same as in February, the Commerce Department reported Friday.
That was above the 2.7% estimate from the Dow Jones consensus.
Including food and energy, the all-items PCE price gauge increased 2.7%, compared with the 2.6% estimate.
On a monthly basis, both measures increased 0.3%, as expected and equaling the increase from February.
Markets showed little reaction to the data, with Wall Street poised to open higher.
Treasury yields fell, with the benchmark 10-year note at 4.67%, down about 0.4 percentage points on the session.
Futures traders grew slightly more optimistic about two potential rate cuts this year, raising the probability to 44%, according to the CME Group’s FedWatch gauge.
“Inflation reports released this morning were not as a hot as feared, but investors should not get overly anchored to the idea that inflation has been completely cured and the Fed will be cutting interest rates in the near-term,” said George Mateyo, chief investment officer at Key Wealth.
“The prospects of rate cuts remain, but they are not assured, and the Fed will likely need weakness in the labor market before they have the confidence to cut.”
Consumers showed that they are still spending despite the elevated price levels.
Personal spending rose 0.8% on the month, a touch higher even than the 0.7% estimate though the same as February.
Personal income increased 0.5%, in line with expectations and higher than the 0.3% increase the previous month.
The personal saving rate fell to 3.2%, down 0.4 percentage points from February and 2 full percentage points from a year ago as households dipped into savings to keep spending afloat.
The report follows bad inflation news from Thursday and likely locks the Fed into holding the line on interest rates through at least the summer unless there is some substantial change in the data.
The Commerce Department reported Thursday that PCE in the first quarter accelerated at a 3.4% annualized rate while gross domestic product increased just 1.6%, well below Wall Street expectations.
With inflation still percolating two years after it began its initial ascent to the highest level in more than 40 years, central bank policymakers are watching the data even more intently as they contemplate the next moves for monetary policy.
The Fed targets 2% inflation, a level that the core PCE has been above for the past three years.
The Fed watches the PCE in particular because it adjusts for changes in consumer behavior and places less weight on housing costs than the more widely circulated consumer price index from the Labor Department.
While they watch both headline and core measures, Fed officials believe the index excluding food and energy provides a better look at longer-run trends as those two categories tend to be more volatile.
Services prices increased 0.4% on the month while goods were up 0.1%, reflecting a swing in consumer prices as goods inflation dominated since the early days of the Covid pandemic.
Food prices showed a 0.1% decline on the month while energy rose 1.2%.
On a 12-month basis, services prices are up 4% while goods have barely moved, increasing just 0.1%.
Food is up 1.5% while energy has gained 2.6%.
https://www.cnbc.com/2024/04/26/pce-inf ... rcent.html
"Key Fed inflation measure rose 2.8% in March from a year ago, more than expected"
Jeff Cox @JEFF.COX.7528 @JEFFCOXCNBCCOM
PUBLISHED FRI, APR 26 2024
KEY POINTS
* The core personal consumption expenditures price index excluding food and energy increased 2.8% from a year ago in March, unchanged from February and slightly higher than expected.
* Personal spending rose 0.8% on the month, more than the personal income increase of 0.5%.
* The personal saving rate fell to 3.2%, down 0.4 percentage points from February and 2 full percentage points from a year ago.
Inflation showed few signs of letting up in March, with a key barometer the Federal Reserve watches closely showing that price pressures remain elevated.
The personal consumption expenditures price index excluding food and energy increased 2.8% from a year ago in March, the same as in February, the Commerce Department reported Friday.
That was above the 2.7% estimate from the Dow Jones consensus.
Including food and energy, the all-items PCE price gauge increased 2.7%, compared with the 2.6% estimate.
On a monthly basis, both measures increased 0.3%, as expected and equaling the increase from February.
Markets showed little reaction to the data, with Wall Street poised to open higher.
Treasury yields fell, with the benchmark 10-year note at 4.67%, down about 0.4 percentage points on the session.
Futures traders grew slightly more optimistic about two potential rate cuts this year, raising the probability to 44%, according to the CME Group’s FedWatch gauge.
“Inflation reports released this morning were not as a hot as feared, but investors should not get overly anchored to the idea that inflation has been completely cured and the Fed will be cutting interest rates in the near-term,” said George Mateyo, chief investment officer at Key Wealth.
“The prospects of rate cuts remain, but they are not assured, and the Fed will likely need weakness in the labor market before they have the confidence to cut.”
Consumers showed that they are still spending despite the elevated price levels.
Personal spending rose 0.8% on the month, a touch higher even than the 0.7% estimate though the same as February.
Personal income increased 0.5%, in line with expectations and higher than the 0.3% increase the previous month.
The personal saving rate fell to 3.2%, down 0.4 percentage points from February and 2 full percentage points from a year ago as households dipped into savings to keep spending afloat.
The report follows bad inflation news from Thursday and likely locks the Fed into holding the line on interest rates through at least the summer unless there is some substantial change in the data.
The Commerce Department reported Thursday that PCE in the first quarter accelerated at a 3.4% annualized rate while gross domestic product increased just 1.6%, well below Wall Street expectations.
With inflation still percolating two years after it began its initial ascent to the highest level in more than 40 years, central bank policymakers are watching the data even more intently as they contemplate the next moves for monetary policy.
The Fed targets 2% inflation, a level that the core PCE has been above for the past three years.
The Fed watches the PCE in particular because it adjusts for changes in consumer behavior and places less weight on housing costs than the more widely circulated consumer price index from the Labor Department.
While they watch both headline and core measures, Fed officials believe the index excluding food and energy provides a better look at longer-run trends as those two categories tend to be more volatile.
Services prices increased 0.4% on the month while goods were up 0.1%, reflecting a swing in consumer prices as goods inflation dominated since the early days of the Covid pandemic.
Food prices showed a 0.1% decline on the month while energy rose 1.2%.
On a 12-month basis, services prices are up 4% while goods have barely moved, increasing just 0.1%.
Food is up 1.5% while energy has gained 2.6%.
https://www.cnbc.com/2024/04/26/pce-inf ... rcent.html
Re: POLITICS
REUTERS
"Investors brace for 5% Treasury yields as US inflation worries mount"
By David Randall
April 26, 2024
NEW YORK, April 26 (Reuters) - As U.S. inflation worries grow, some investors are preparing for the 10-year U.S. Treasury yield to breach a 16-year high of 5% hit last October.
Bond yields, which move inversely to prices, have climbed in recent weeks as signs of persistent inflation erode expectations for how deeply the Federal Reserve will be able to cut interest rates without further fueling consumer prices.
The yield on the benchmark 10-year note is up 80 basis points this year and last stood at 4.70%, a five-month high.
Many investors are betting further weakness lies ahead for bonds.
Global fund managers' fixed income allocations in the latest BofA Global Research survey are down to their lowest level since 2003.
Bearish Treasury positioning among some classes of hedge funds stands at its highest level of the year, according to BofA data, even as other asset managers have increased their bullish bets.
"It all boils down to one word: inflation."
"If the market doesn't see signs that inflation is contained, then there's no reason that yields won't keep pushing higher," said Don Ellenberger, senior portfolio manager at Federated Hermes.
He has decreased his portfolio's interest rate sensitivity, wary that sticky inflation and labor market strength could push yields as high as 5.25%.
Further evidence that inflation is heating up again came on Thursday, with data showing the personal consumption expenditures (PCE) price index excluding food and energy rose far more than expected in the first quarter.
Futures markets showed investors now expect the Fed to deliver just 35 basis points in rate cuts this year, compared to the more than 150 points that were priced in at the beginning of 2024.
Another hot inflation reading on Friday, when PCE data for March will be released, could further close the window on rate-cut expectations this year.
More insights on the economy could come at the conclusion of the U.S. central bank's monetary policy meeting on May 1.
'HIGH-WATER MARK'
The level of Treasury yields is closely watched by market participants, as elevated yields can translate into higher borrowing costs for consumers and companies and tighten financial conditions in the economy.
A sharp run-up in yields during the latter part of 2023 sparked a sell-off in the S&P 500, though equities rebounded when yields reversed.
This year's rally in stocks has stumbled in recent weeks as yields have risen, with the S&P 500 cutting its gains to around 6% on a year-to-date basis, from more than 10%.
Some investors have used the weakness in bonds to add to their fixed income holdings, confident that yields are unlikely to rise much further unless the Fed says it is looking to once again raise its benchmark overnight interest rate from the current 5.25%-5.50% range.
Others, however, have been skeptical inflation will cool anytime soon.
"Inflation is not coming down like the Fed thought it was," said Arthur Laffer, president of Laffer Tengler Investments, who is bearish on longer-dated Treasuries and believes yields could rise as high as 6%.
"You're not getting paid to take risk in the bond market right now."
Michael Purves, head of Tallbacken Capital Advisors, wrote it's "not inconceivable" that the 10-year Treasury yield could reach its 2007 high of 5.22%, if higher prices for oil and other raw materials continue pushing up inflation.
The price of Brent crude is up about 17% on a year-to-date basis, even after retreating in the last week on easing fears of a wider conflict in the Middle East.
Fiscal worries are another factor that could push yields higher.
Ratings agency Fitch downgraded the U.S. credit rating last year partly due to concern over rising debt levels.
Many investors anticipate a rise in term premiums - or the compensation demanded to hold long-term debt.
"The fiscal conditions of the U.S. are starting to matter, and it can put tremendous pressure on yields and push down on equity valuations in a very short period of time if the market starts to worry more," said Bryant VanCronkhite, a senior portfolio manager at Allspring Global Investments, who expects 10-year Treasury yields to move above 5%.
Still, there are reasons to think a return to 5% yields would be a "high-water mark" for investors, said Alex Christensen, a portfolio manager at Columbia Threadneedle Investments who is overweight two-year Treasuries.
The market narrative that dominated since the so-called Fed pivot in December "was very one-sided and left little room for changes in the inflationary trend," Christensen said.
He believes the Fed is unlikely to pivot towards rate increases.
"We think the general inflationary trend is stable to lower," he said.
Reporting by David Randall; Editing by Ira Iosebashvili and Paul Simao
https://www.reuters.com/markets/us/inve ... 024-04-26/
"Investors brace for 5% Treasury yields as US inflation worries mount"
By David Randall
April 26, 2024
NEW YORK, April 26 (Reuters) - As U.S. inflation worries grow, some investors are preparing for the 10-year U.S. Treasury yield to breach a 16-year high of 5% hit last October.
Bond yields, which move inversely to prices, have climbed in recent weeks as signs of persistent inflation erode expectations for how deeply the Federal Reserve will be able to cut interest rates without further fueling consumer prices.
The yield on the benchmark 10-year note is up 80 basis points this year and last stood at 4.70%, a five-month high.
Many investors are betting further weakness lies ahead for bonds.
Global fund managers' fixed income allocations in the latest BofA Global Research survey are down to their lowest level since 2003.
Bearish Treasury positioning among some classes of hedge funds stands at its highest level of the year, according to BofA data, even as other asset managers have increased their bullish bets.
"It all boils down to one word: inflation."
"If the market doesn't see signs that inflation is contained, then there's no reason that yields won't keep pushing higher," said Don Ellenberger, senior portfolio manager at Federated Hermes.
He has decreased his portfolio's interest rate sensitivity, wary that sticky inflation and labor market strength could push yields as high as 5.25%.
Further evidence that inflation is heating up again came on Thursday, with data showing the personal consumption expenditures (PCE) price index excluding food and energy rose far more than expected in the first quarter.
Futures markets showed investors now expect the Fed to deliver just 35 basis points in rate cuts this year, compared to the more than 150 points that were priced in at the beginning of 2024.
Another hot inflation reading on Friday, when PCE data for March will be released, could further close the window on rate-cut expectations this year.
More insights on the economy could come at the conclusion of the U.S. central bank's monetary policy meeting on May 1.
'HIGH-WATER MARK'
The level of Treasury yields is closely watched by market participants, as elevated yields can translate into higher borrowing costs for consumers and companies and tighten financial conditions in the economy.
A sharp run-up in yields during the latter part of 2023 sparked a sell-off in the S&P 500, though equities rebounded when yields reversed.
This year's rally in stocks has stumbled in recent weeks as yields have risen, with the S&P 500 cutting its gains to around 6% on a year-to-date basis, from more than 10%.
Some investors have used the weakness in bonds to add to their fixed income holdings, confident that yields are unlikely to rise much further unless the Fed says it is looking to once again raise its benchmark overnight interest rate from the current 5.25%-5.50% range.
Others, however, have been skeptical inflation will cool anytime soon.
"Inflation is not coming down like the Fed thought it was," said Arthur Laffer, president of Laffer Tengler Investments, who is bearish on longer-dated Treasuries and believes yields could rise as high as 6%.
"You're not getting paid to take risk in the bond market right now."
Michael Purves, head of Tallbacken Capital Advisors, wrote it's "not inconceivable" that the 10-year Treasury yield could reach its 2007 high of 5.22%, if higher prices for oil and other raw materials continue pushing up inflation.
The price of Brent crude is up about 17% on a year-to-date basis, even after retreating in the last week on easing fears of a wider conflict in the Middle East.
Fiscal worries are another factor that could push yields higher.
Ratings agency Fitch downgraded the U.S. credit rating last year partly due to concern over rising debt levels.
Many investors anticipate a rise in term premiums - or the compensation demanded to hold long-term debt.
"The fiscal conditions of the U.S. are starting to matter, and it can put tremendous pressure on yields and push down on equity valuations in a very short period of time if the market starts to worry more," said Bryant VanCronkhite, a senior portfolio manager at Allspring Global Investments, who expects 10-year Treasury yields to move above 5%.
Still, there are reasons to think a return to 5% yields would be a "high-water mark" for investors, said Alex Christensen, a portfolio manager at Columbia Threadneedle Investments who is overweight two-year Treasuries.
The market narrative that dominated since the so-called Fed pivot in December "was very one-sided and left little room for changes in the inflationary trend," Christensen said.
He believes the Fed is unlikely to pivot towards rate increases.
"We think the general inflationary trend is stable to lower," he said.
Reporting by David Randall; Editing by Ira Iosebashvili and Paul Simao
https://www.reuters.com/markets/us/inve ... 024-04-26/
Re: POLITICS
REUTERS
"Chipmaker Intel falls as AI competition hurts forecast"
By Reuters
April 26, 2024
April 26 (Reuters) - Intel shares slumped more than 12% on Friday after a downbeat forecast signaled that the boom in AI was diverting enterprise spending away from its traditional data center chips.
The stock has fallen around 30% so far this year as Intel trails rival chip companies such as Nvidia in producing advanced artificial intelligence (AI) chips and components.
Intel forecast second-quarter revenue of $12.5 billion to $13.5 billion, compared with analysts' average estimate of $13.57 billion, according to LSEG data.
"While we believe they are doing everything they can to try to repair things, it is clear the company is profoundly broken, and it will take years to see the fruits of their (currently exhaustive) labor," Bernstein analysts said in a note.
Intel is planning a $100 billion spending spree across four U.S. states to build and expand factories.
It also unveiled a new AI chip earlier this year to keep up with competition.
Friday's drop was set to erase nearly $19 billion from the company's market value, which stood at $149.4 billion as of Thursday's close.
Businesses have prioritized spending on advanced and speedy AI server chips, hurting demand for Intel's central processing units, which had long been the mainstay chip powering data centers.
Although encouraged by the launch of Intel's Gaudi 3 AI chip, "we worry the company will continue to cede wallet share within the overall data center compute market to the likes of Nvidia and Arm", Goldman Sachs analysts said.
Still, Intel is optimistic that a fresh upgrade cycle for personal computers around a new version of Microsoft's Windows operating system will help PC sales in the second half of the year.
That could translate to more demand for its chips used in those devices.
The company's earnings contrasted strong results from Microsoft and Alphabet, which are Nvidia clients and also design in-house chips for their data centers.
Reporting by Zaheer Kachwala in Bengaluru; Editing by Devika Syamnath
https://www.reuters.com/technology/inte ... 024-04-26/
"Chipmaker Intel falls as AI competition hurts forecast"
By Reuters
April 26, 2024
April 26 (Reuters) - Intel shares slumped more than 12% on Friday after a downbeat forecast signaled that the boom in AI was diverting enterprise spending away from its traditional data center chips.
The stock has fallen around 30% so far this year as Intel trails rival chip companies such as Nvidia in producing advanced artificial intelligence (AI) chips and components.
Intel forecast second-quarter revenue of $12.5 billion to $13.5 billion, compared with analysts' average estimate of $13.57 billion, according to LSEG data.
"While we believe they are doing everything they can to try to repair things, it is clear the company is profoundly broken, and it will take years to see the fruits of their (currently exhaustive) labor," Bernstein analysts said in a note.
Intel is planning a $100 billion spending spree across four U.S. states to build and expand factories.
It also unveiled a new AI chip earlier this year to keep up with competition.
Friday's drop was set to erase nearly $19 billion from the company's market value, which stood at $149.4 billion as of Thursday's close.
Businesses have prioritized spending on advanced and speedy AI server chips, hurting demand for Intel's central processing units, which had long been the mainstay chip powering data centers.
Although encouraged by the launch of Intel's Gaudi 3 AI chip, "we worry the company will continue to cede wallet share within the overall data center compute market to the likes of Nvidia and Arm", Goldman Sachs analysts said.
Still, Intel is optimistic that a fresh upgrade cycle for personal computers around a new version of Microsoft's Windows operating system will help PC sales in the second half of the year.
That could translate to more demand for its chips used in those devices.
The company's earnings contrasted strong results from Microsoft and Alphabet, which are Nvidia clients and also design in-house chips for their data centers.
Reporting by Zaheer Kachwala in Bengaluru; Editing by Devika Syamnath
https://www.reuters.com/technology/inte ... 024-04-26/
Re: POLITICS
ABC News
Trump leads in swing-state polls and is tied with Biden nationally"
26 APRIL 2023
It's the moment you've all been waiting for: Today, 538 launched our interactive polling averages for the 2024 presidential general election.
They show incumbent President Joe Biden and former President Donald Trump essentially tied in national polls and Trump with a tenuous lead in key swing states.
Specifically, as of April 25 at 9 a.m. Eastern, our national polling average shows 40.9 percent of likely voters would support Biden if the election were held today, while 41.6 percent would support Trump.
But there's considerable uncertainty in those numbers.
To communicate this, we are also publishing uncertainty intervals for our horse-race averages for the first time.
These intervals — represented by the red and blue shaded areas around each line — are kind of like the range of possible precipitation reported in a weather forecast, showing you could get anywhere from, say, 1 to 3 inches of rain in an upcoming storm.
Our uncertainty intervals take into account the variability of the polling data and the uncertainty we have about the various adjustments we are making, which are detailed later in this article.
Right now, that interval shows that Biden's support could be anywhere between 39.4 and 42.2 percent, while Trump’s range is from 40.3 to 42.8 percent.
Read on for more information about all the sources of uncertainty we are (and aren't) taking into account for these averages.
Our averages also show Trump leading in most swing states, though there is enough uncertainty that Biden could easily be ahead in enough to win the Electoral College.
The table below shows each candidate's current estimated support and our uncertainty intervals for those numbers.
Of course, the election will not be held today.
While 538's new way of calculating polling averages takes into account sources of error that could affect those averages today — such as having a lot of polls from pollsters who tend to favor one party over another — we make no attempt to account for error that could arise from the large amount of time (six months!) remaining until the election, nor do we include the chance of an industrywide polling miss favoring either candidate.
Those are the types of potential errors we will account for in our election forecast, which will use several ingredients (including these averages) to predict the outcome of the November election and which we'll release in a couple of months.
By contrast, our polling averages are intended to be our best guess at where public opinion stands right now, not a prediction for an election happening several months from now.
We still urge you to exercise caution when reading early polls; they can change significantly between now and Election Day.
The rest of this article explains the methodology behind our polling averages: which polls we collect, how we weight and adjust them, and how we average them.
This methodology is different from the one powering our other polling averages (such as for Biden's presidential job approval rating) in a few important ways, so it's worth spending a little time on the details.
Which polls we include
The building blocks of our polling averages are, of course, individual polls.
Our presidential general election polling averages include all publicly available scientific polls that meet our methodological and ethical standards and test at least Biden versus Trump, though we prefer polls that test all major candidates who will be on the ballot in most states.
We consider the Democratic and Republican presidential nominees to be "major" candidates by default, and include in that category any third-party candidate that is polling at least in the mid single digits nationally and will be on the ballot in most states, or who has a sizable chance of coming in second place in any state (think Evan McMullin in 2016).
That means we exclude polls that ask people how they would vote in hypothetical matchups, like if Michelle Obama were the Democratic nominee instead of Biden.
We also don't use polls that ask about support for a "generic Democrat" or "generic Republican," nor do we include polls that ask voters if they support "someone else" for a third-party option.
If a pollster releases versions of a survey among multiple populations, our horse-race averages use the one with the most restrictive sample.
That means we prefer polls of likely voters to polls of registered voters and polls of registered voters to polls of all adults.
If a pollster releases multiple likely voter estimates, we average them together.
Some pollsters conduct polls that overlap with each other, such as Morning Consult.
In these so-called "tracking polls," interviews from a certain period are incorporated into future polls, getting reweighted with different samples until they are too old and dropped from the analysis.
But our averages include only polls with non-overlapping dates.
How this works in practice is that we include the most recent iteration of each tracking poll, then include the next most recent version of that poll that doesn't overlap at all with the first poll, then include the next most recent version that doesn't overlap with that poll, etc.
Finally, to answer a frequently asked question, we do include partisan and internal campaign polls that get released to the public, with adjustments that aim to cancel out their partisan biases (see below).
These polls make up a fairly small percentage of the data 538 aggregates, but they can still be useful for calculating trends in support over the course of the campaign.
How we weight polls
The next step is to calculate how much weight each poll should be given in our average.
For really cool, but wonky, reasons you can read about in the footnotes,* our new presidential general election polling averages don't need to weight polls by several factors that we previously weighted on.
Instead, a poll's weight is now a function of just two factors.
(While our model does not weight for these old factors, they still get taken into account. See the footnote.)
The first is the 538 rating of the pollster that conducted the poll.**
These ratings distill the empirical record and methodological transparency of each pollster into a single rating (from 0.5 to 3.0 stars) that tells our model how seriously to take its polls.
Each poll gets a weight equal to the square root of its pollster's rating divided by the maximum three stars, then that weight is divided by the weight of the average pollster.
The resulting pollster weight on a typical day usually ranges between a maximum of 1.3 and minimum of 0.5.
If a pollster is new and does not have a 538 pollster rating, its polls get a weight equal to one.
Polls also receive reduced weight if the pollster who conducted them has released a lot of surveys in a short period of time.
Specifically, each poll gets a weight equal to the square root of the number of polls released by its pollster in a 14-day window.
We make this adjustment to prevent any given pollster from exerting too much control over the average.
Finally, we no longer explicitly weight polls by sample size, but we do limit the effective number of interviews per poll.
Some polls contain tens of thousands of interviews (usually done very cheaply), and leaving these large numbers as they are would give those polls too much weight in our average.
Therefore, we cap sample sizes at 5,000.
How we adjust and average polls
Let's zoom out for a second.
Broadly speaking, the most commonly used polling averages for U.S. elections have followed one of three approaches:
* Take a simple moving average of polls released over some number of previous days (the RealClearPolitics approach).
* Calculate a trendline through the polls using various statistical techniques, such as a polynomial trendline or Kalman filter (the HuffPost Pollster and The Economist approach).
* Combine these approaches, putting a certain amount of weight on the moving average and the rest on the fancier trend (historically, this has been the 538 approach).
There are a lot of benefits to this third option.
The average-of-averages approach allows you to use predictions from the best parts of a slow-moving exponentially weighted moving average and a fast-moving polynomial trendline; it is computationally efficient to do so; and it's easy to explain this model to the public.
It's also relatively trivial to tweak the model if we find something is working incorrectly.
However, this model has some shortcomings too.
Our poll-averaging model for approval rating, favorability rating, primary elections and the generic congressional ballot is really a set of many different models that work together iteratively: First, we use models to reduce the weight on polls that are very old or have small sample sizes; then we use models to average polls and detect outliers; then we run new averaging models to detect house effects; and so on and so on, for nearly a dozen individual steps.
This can introduce some problems, some of them practical and others statistical, if a modeler is not careful.
First, it's hard to account for uncertainty in the average, especially when using ad hoc weights for sample size and other factors.
That's because we generate potential statistical error every time we move from one model to the next, and we have to run the program thousands of times every time we want to update!
It's also a little more sensitive to noise than we'd like it to be, even when designed to accurately predict support for candidates in previous campaigns.
So this year, we're unveiling a brand-new methodology for averaging polls of the presidential general election.
While the general approach is similar — we take polls, we weight them, we adjust them and we average them together — the various steps of our model no longer happen iteratively, but rather simultaneously in the same combined model.
Our new model is formally called a Bayesian multi-level dynamic linear model, or a "state-space" model depending on your persuasion.
The model is fit using a statistical method called Markov chain Monte Carlo — a mathematical approach to figuring out the likely values of a bunch of different parameters in one huge equation.
Oversimplifying a bit, you can think of our presidential general-election polling average as one giant model that is trying to predict the results of polls we collect based on (1) the overall state of public opinion on any given day and (2) various factors that could have influenced the result of a particular poll.
For example: The polling firm responsible for the poll: The specific ways in which a pollster collects and analyzes its data can lead to systematic differences between its polls and other pollsters' polls.
For example, some pollsters, such as the Trafalgar Group, usually underestimate support for Democrats, while other pollsters, like Center Street PAC, overestimate them.
We call these differences "house effects," and we apply a house-effect adjustment to ensure they're not biasing our averages.
The mode used to conduct the poll: Different groups of people answer polls in different ways.
Young people are likelier to be online, for instance, and phone polls reach older, whiter voters more readily.
If pollsters aren't careful, these biases can creep into all polls conducted using a single mode.
So we apply a mode-effects adjustment to correct for those biases before we aggregate those surveys.
Whether the poll sampled likely voters, registered voters or all adults: Our presidential general-election averages apply a likely-voter adjustment to all polls of registered voters and adults, since, at the end of the day, we're not interested in Trump's and Biden's support among people who aren't going to turn out.
Whether the poll included response options for third-party candidates or forced respondents to pick between Biden and Trump: This year, independent candidate Robert F. Kennedy Jr. is likely to make the ballot in most, if not all states, and since he's polling at around 10 percent according to our average, polls that don't include him as an option are likely overestimating support for Biden and Trump.
So we apply a third-party adjustment to polls that don't ask about third parties.
Whether the poll was conducted by a campaign or other partisan organization: We apply a partisanship adjustment to account for this.
Our initial guess is partisan polls overestimate support for the allied candidate by 2 percentage points, but this effect will shrink (or grow) if the data reveals a different answer.
Our prediction for a given poll also accounts for the value of the polling average on the day it was conducted.
That's because if overall support for a candidate is 50 percent, we should expect polls from that day to reveal higher support than if they were at, say, 30 percent overall support.
This also means the model implicitly puts less weight on polls that look like huge outliers, after adjusting for all the factors above.
At the state level, we consider one more variable: the national political environment.
Historically, candidates' support has tended to rise and fall across the country more or less uniformly.
In other words, if a candidate slips in national polls, they have probably slipped in state polls by a similar amount (our nation is made up of the states, after all).
Our trendline adjustment takes this into account.
Every day, movement in a given state's polling average is a function of changes in both polls of that state and national polls.
The relationship also works the other way — if we see a flood of state polls with Biden improving by 3 points, his national average will increase, too.
But if our prediction for the support of Candidate X in State Y on Day Z depends on the average for that X-Y-Z combination, that brings up the question of how exactly the national and state averages are being calculated.
We use a random walk to model averages over time.
In essence, we tell our computers that support for every candidate in state and national polls should start at some point on Day 1 and move by some amount on average each subsequent day.
Support for Biden might move by 0.1 points on Day 2, -0.2 points on Day 3, 0.4 points on Day 4, 0 points on Day 5, and so on and so on.
Every time we run our model, it determines the likeliest values of these daily changes for each candidate nationally and in each state, while adjusting polls for all the factors mentioned above.
We also account for the amount of random error in each poll.
This is noise that goes above and beyond the patterns of bias we can account for with the adjustments listed above.
The primary source of this noise is sampling error, derived from the number of interviews a pollster does: A larger sample size means less variance due to "stochasticity" (random weirdness) in a poll's sample.
But there is also non-sampling error in each poll — a blanket term encompassing any additional noise that could be a result of faulty weights, certain groups not picking up the phone, a bad questionnaire design or really anything else that we haven't added explicit adjustment factors for.
Our model decides how much non-sampling error is present across the polls by adding an additional constant to the standard deviation implied by each poll's sample size via the sum of squares formula (with the model deciding how large that constant should be).
Considering all these factors together, the full equation behind our model looks like this.
Two more short notes: First, we also tell the model to expect support for one candidate to move inversely with support for the other.
If Biden makes gains in Wisconsin, it's probably coming at Trump's expense.
While this correlation is more important for forecasting outcomes than averaging polls (it changes average support for Biden by just 0.1 points today), we find that our models run much faster if they take this into account.
The exact correlation between the candidates is also a parameter in the model; we start with a strong prior that extra Democratic votes cost the Republicans (and vice versa), but the data will determine how strong the penalty is.
Second, our model no longer directly measures each state's "elasticity," or how malleable its public opinion is compared with the nation as a whole.
However, the model implicitly accounts for this by letting the day-to-day variability in candidates' averages change by different amounts in different states.
How we account for uncertainty
Our new model also lets us account for uncertainty in the polling average in a very straightforward way.
Imagine we are not calculating support for Biden in Wisconsin one single time, but thousands of times, where each time we see what his support would be if the data and parameters of our model had different values.
What if the latest poll from The New York Times/Siena College had Biden's support 3 points higher, as a poll of 1,000 people would be expected to have about 5 percent of the time?
Or what if the third-party adjustment were smaller?
Our model answers these questions by simulating thousands of different polling averages each time it runs.
That, in turn, lets us show uncertainty intervals directly on the average — to our knowledge, the first time a general-election polling average has done so since HuffPost Pollster in 2016.
Again, though, these uncertainty intervals aren't showing you the uncertainty that arises from the possibility of an industrywide polling miss.
That is the job of our forecast.
The uncertainty in our averages answers the question, "How different could support for each candidate be in the average poll today if we observed more polls with slightly different results?", while the uncertainty in our forecast answers the question, "How different could each candidate's actual support be from the polls?"
Conclusion
I should note that variations of this model have been proposed by political scientists and statisticians for a few decades.
(The following paragraph is not exhaustive, but should give you a good sense of the history of this approach.)
By my research, Robert Erikson and Christopher Wlezien were the first to model support for candidates with house effects and as a smooth function of time over the campaign.
Political scientist Simon Jackman later formalized a Bayesian time-series model of polls with house effects for Australian elections.
Statisticians Kari Lock and Andrew Gelman employed a time-series model of polls as part of a paper forecasting election outcomes with polls and other data.
Political scientist/survey statistician/pollster Drew Linzer combined aspects of these and other approaches in a Bayesian dynamic linear model of state and national polls of 2012 general election.
And I worked with Gelman and statistician Merlin Heidemanns to add additional poll-level adjustments as well as other factors to model the 2020 general election in this way.
538's new presidential general-election polling average goes one step beyond these previous approaches by modeling support for multiple candidates across geographies simultaneously and adding a few additional adjustments that have not been attempted before in the fully Bayesian context.
That's it for now!
If I've done my job well, hopefully you can see how this new approach closely resembles our original dual-purpose model of models, but in one neat package.
If you see any polls that are missing, or spot a methodological bug (these things do happen when we launch new averages, especially with models of this level of sophistication), hit us up with an email.
Footnotes
*More on this in the "How we adjust and average polls" section, but the basic reason is that our new model is not a mix of a weighted average and polynomial trendline — methods that need to be told explicitly which observations are subject to less noise — but rather a Bayesian time series model that directly takes the observation noise in the poll into account.
So we no longer need to weight on sample size, poll recency and the degree to which a poll is an outlier; the model handles that stuff for us on its own.
**As a methodological matter, we think how to best integrate pollster weights into an aggregation model is an open debate.
(In fact, one statistician told me we shouldn't weight to pollster effects at all — instead we should create complicated meta-models of polling accuracy across many different contests simultaneously and inject the estimated uncertainty and bias of each pollster directly into our model. Unfortunately, this is not computationally feasible for the type of live-updating statistical model we want to publish.)
We tried a few more complicated methods — such as adjusting the effective sample size of each poll to match how accurate we expected it to be — but they did not perform meaningfully better than more parsimonious approaches.
As a practical matter, what matters more than how you weight polls based on their quality is that you try to account for quality at all.
https://www.msn.com/en-us/news/politics ... 418a&ei=36
Trump leads in swing-state polls and is tied with Biden nationally"
26 APRIL 2023
It's the moment you've all been waiting for: Today, 538 launched our interactive polling averages for the 2024 presidential general election.
They show incumbent President Joe Biden and former President Donald Trump essentially tied in national polls and Trump with a tenuous lead in key swing states.
Specifically, as of April 25 at 9 a.m. Eastern, our national polling average shows 40.9 percent of likely voters would support Biden if the election were held today, while 41.6 percent would support Trump.
But there's considerable uncertainty in those numbers.
To communicate this, we are also publishing uncertainty intervals for our horse-race averages for the first time.
These intervals — represented by the red and blue shaded areas around each line — are kind of like the range of possible precipitation reported in a weather forecast, showing you could get anywhere from, say, 1 to 3 inches of rain in an upcoming storm.
Our uncertainty intervals take into account the variability of the polling data and the uncertainty we have about the various adjustments we are making, which are detailed later in this article.
Right now, that interval shows that Biden's support could be anywhere between 39.4 and 42.2 percent, while Trump’s range is from 40.3 to 42.8 percent.
Read on for more information about all the sources of uncertainty we are (and aren't) taking into account for these averages.
Our averages also show Trump leading in most swing states, though there is enough uncertainty that Biden could easily be ahead in enough to win the Electoral College.
The table below shows each candidate's current estimated support and our uncertainty intervals for those numbers.
Of course, the election will not be held today.
While 538's new way of calculating polling averages takes into account sources of error that could affect those averages today — such as having a lot of polls from pollsters who tend to favor one party over another — we make no attempt to account for error that could arise from the large amount of time (six months!) remaining until the election, nor do we include the chance of an industrywide polling miss favoring either candidate.
Those are the types of potential errors we will account for in our election forecast, which will use several ingredients (including these averages) to predict the outcome of the November election and which we'll release in a couple of months.
By contrast, our polling averages are intended to be our best guess at where public opinion stands right now, not a prediction for an election happening several months from now.
We still urge you to exercise caution when reading early polls; they can change significantly between now and Election Day.
The rest of this article explains the methodology behind our polling averages: which polls we collect, how we weight and adjust them, and how we average them.
This methodology is different from the one powering our other polling averages (such as for Biden's presidential job approval rating) in a few important ways, so it's worth spending a little time on the details.
Which polls we include
The building blocks of our polling averages are, of course, individual polls.
Our presidential general election polling averages include all publicly available scientific polls that meet our methodological and ethical standards and test at least Biden versus Trump, though we prefer polls that test all major candidates who will be on the ballot in most states.
We consider the Democratic and Republican presidential nominees to be "major" candidates by default, and include in that category any third-party candidate that is polling at least in the mid single digits nationally and will be on the ballot in most states, or who has a sizable chance of coming in second place in any state (think Evan McMullin in 2016).
That means we exclude polls that ask people how they would vote in hypothetical matchups, like if Michelle Obama were the Democratic nominee instead of Biden.
We also don't use polls that ask about support for a "generic Democrat" or "generic Republican," nor do we include polls that ask voters if they support "someone else" for a third-party option.
If a pollster releases versions of a survey among multiple populations, our horse-race averages use the one with the most restrictive sample.
That means we prefer polls of likely voters to polls of registered voters and polls of registered voters to polls of all adults.
If a pollster releases multiple likely voter estimates, we average them together.
Some pollsters conduct polls that overlap with each other, such as Morning Consult.
In these so-called "tracking polls," interviews from a certain period are incorporated into future polls, getting reweighted with different samples until they are too old and dropped from the analysis.
But our averages include only polls with non-overlapping dates.
How this works in practice is that we include the most recent iteration of each tracking poll, then include the next most recent version of that poll that doesn't overlap at all with the first poll, then include the next most recent version that doesn't overlap with that poll, etc.
Finally, to answer a frequently asked question, we do include partisan and internal campaign polls that get released to the public, with adjustments that aim to cancel out their partisan biases (see below).
These polls make up a fairly small percentage of the data 538 aggregates, but they can still be useful for calculating trends in support over the course of the campaign.
How we weight polls
The next step is to calculate how much weight each poll should be given in our average.
For really cool, but wonky, reasons you can read about in the footnotes,* our new presidential general election polling averages don't need to weight polls by several factors that we previously weighted on.
Instead, a poll's weight is now a function of just two factors.
(While our model does not weight for these old factors, they still get taken into account. See the footnote.)
The first is the 538 rating of the pollster that conducted the poll.**
These ratings distill the empirical record and methodological transparency of each pollster into a single rating (from 0.5 to 3.0 stars) that tells our model how seriously to take its polls.
Each poll gets a weight equal to the square root of its pollster's rating divided by the maximum three stars, then that weight is divided by the weight of the average pollster.
The resulting pollster weight on a typical day usually ranges between a maximum of 1.3 and minimum of 0.5.
If a pollster is new and does not have a 538 pollster rating, its polls get a weight equal to one.
Polls also receive reduced weight if the pollster who conducted them has released a lot of surveys in a short period of time.
Specifically, each poll gets a weight equal to the square root of the number of polls released by its pollster in a 14-day window.
We make this adjustment to prevent any given pollster from exerting too much control over the average.
Finally, we no longer explicitly weight polls by sample size, but we do limit the effective number of interviews per poll.
Some polls contain tens of thousands of interviews (usually done very cheaply), and leaving these large numbers as they are would give those polls too much weight in our average.
Therefore, we cap sample sizes at 5,000.
How we adjust and average polls
Let's zoom out for a second.
Broadly speaking, the most commonly used polling averages for U.S. elections have followed one of three approaches:
* Take a simple moving average of polls released over some number of previous days (the RealClearPolitics approach).
* Calculate a trendline through the polls using various statistical techniques, such as a polynomial trendline or Kalman filter (the HuffPost Pollster and The Economist approach).
* Combine these approaches, putting a certain amount of weight on the moving average and the rest on the fancier trend (historically, this has been the 538 approach).
There are a lot of benefits to this third option.
The average-of-averages approach allows you to use predictions from the best parts of a slow-moving exponentially weighted moving average and a fast-moving polynomial trendline; it is computationally efficient to do so; and it's easy to explain this model to the public.
It's also relatively trivial to tweak the model if we find something is working incorrectly.
However, this model has some shortcomings too.
Our poll-averaging model for approval rating, favorability rating, primary elections and the generic congressional ballot is really a set of many different models that work together iteratively: First, we use models to reduce the weight on polls that are very old or have small sample sizes; then we use models to average polls and detect outliers; then we run new averaging models to detect house effects; and so on and so on, for nearly a dozen individual steps.
This can introduce some problems, some of them practical and others statistical, if a modeler is not careful.
First, it's hard to account for uncertainty in the average, especially when using ad hoc weights for sample size and other factors.
That's because we generate potential statistical error every time we move from one model to the next, and we have to run the program thousands of times every time we want to update!
It's also a little more sensitive to noise than we'd like it to be, even when designed to accurately predict support for candidates in previous campaigns.
So this year, we're unveiling a brand-new methodology for averaging polls of the presidential general election.
While the general approach is similar — we take polls, we weight them, we adjust them and we average them together — the various steps of our model no longer happen iteratively, but rather simultaneously in the same combined model.
Our new model is formally called a Bayesian multi-level dynamic linear model, or a "state-space" model depending on your persuasion.
The model is fit using a statistical method called Markov chain Monte Carlo — a mathematical approach to figuring out the likely values of a bunch of different parameters in one huge equation.
Oversimplifying a bit, you can think of our presidential general-election polling average as one giant model that is trying to predict the results of polls we collect based on (1) the overall state of public opinion on any given day and (2) various factors that could have influenced the result of a particular poll.
For example: The polling firm responsible for the poll: The specific ways in which a pollster collects and analyzes its data can lead to systematic differences between its polls and other pollsters' polls.
For example, some pollsters, such as the Trafalgar Group, usually underestimate support for Democrats, while other pollsters, like Center Street PAC, overestimate them.
We call these differences "house effects," and we apply a house-effect adjustment to ensure they're not biasing our averages.
The mode used to conduct the poll: Different groups of people answer polls in different ways.
Young people are likelier to be online, for instance, and phone polls reach older, whiter voters more readily.
If pollsters aren't careful, these biases can creep into all polls conducted using a single mode.
So we apply a mode-effects adjustment to correct for those biases before we aggregate those surveys.
Whether the poll sampled likely voters, registered voters or all adults: Our presidential general-election averages apply a likely-voter adjustment to all polls of registered voters and adults, since, at the end of the day, we're not interested in Trump's and Biden's support among people who aren't going to turn out.
Whether the poll included response options for third-party candidates or forced respondents to pick between Biden and Trump: This year, independent candidate Robert F. Kennedy Jr. is likely to make the ballot in most, if not all states, and since he's polling at around 10 percent according to our average, polls that don't include him as an option are likely overestimating support for Biden and Trump.
So we apply a third-party adjustment to polls that don't ask about third parties.
Whether the poll was conducted by a campaign or other partisan organization: We apply a partisanship adjustment to account for this.
Our initial guess is partisan polls overestimate support for the allied candidate by 2 percentage points, but this effect will shrink (or grow) if the data reveals a different answer.
Our prediction for a given poll also accounts for the value of the polling average on the day it was conducted.
That's because if overall support for a candidate is 50 percent, we should expect polls from that day to reveal higher support than if they were at, say, 30 percent overall support.
This also means the model implicitly puts less weight on polls that look like huge outliers, after adjusting for all the factors above.
At the state level, we consider one more variable: the national political environment.
Historically, candidates' support has tended to rise and fall across the country more or less uniformly.
In other words, if a candidate slips in national polls, they have probably slipped in state polls by a similar amount (our nation is made up of the states, after all).
Our trendline adjustment takes this into account.
Every day, movement in a given state's polling average is a function of changes in both polls of that state and national polls.
The relationship also works the other way — if we see a flood of state polls with Biden improving by 3 points, his national average will increase, too.
But if our prediction for the support of Candidate X in State Y on Day Z depends on the average for that X-Y-Z combination, that brings up the question of how exactly the national and state averages are being calculated.
We use a random walk to model averages over time.
In essence, we tell our computers that support for every candidate in state and national polls should start at some point on Day 1 and move by some amount on average each subsequent day.
Support for Biden might move by 0.1 points on Day 2, -0.2 points on Day 3, 0.4 points on Day 4, 0 points on Day 5, and so on and so on.
Every time we run our model, it determines the likeliest values of these daily changes for each candidate nationally and in each state, while adjusting polls for all the factors mentioned above.
We also account for the amount of random error in each poll.
This is noise that goes above and beyond the patterns of bias we can account for with the adjustments listed above.
The primary source of this noise is sampling error, derived from the number of interviews a pollster does: A larger sample size means less variance due to "stochasticity" (random weirdness) in a poll's sample.
But there is also non-sampling error in each poll — a blanket term encompassing any additional noise that could be a result of faulty weights, certain groups not picking up the phone, a bad questionnaire design or really anything else that we haven't added explicit adjustment factors for.
Our model decides how much non-sampling error is present across the polls by adding an additional constant to the standard deviation implied by each poll's sample size via the sum of squares formula (with the model deciding how large that constant should be).
Considering all these factors together, the full equation behind our model looks like this.
Two more short notes: First, we also tell the model to expect support for one candidate to move inversely with support for the other.
If Biden makes gains in Wisconsin, it's probably coming at Trump's expense.
While this correlation is more important for forecasting outcomes than averaging polls (it changes average support for Biden by just 0.1 points today), we find that our models run much faster if they take this into account.
The exact correlation between the candidates is also a parameter in the model; we start with a strong prior that extra Democratic votes cost the Republicans (and vice versa), but the data will determine how strong the penalty is.
Second, our model no longer directly measures each state's "elasticity," or how malleable its public opinion is compared with the nation as a whole.
However, the model implicitly accounts for this by letting the day-to-day variability in candidates' averages change by different amounts in different states.
How we account for uncertainty
Our new model also lets us account for uncertainty in the polling average in a very straightforward way.
Imagine we are not calculating support for Biden in Wisconsin one single time, but thousands of times, where each time we see what his support would be if the data and parameters of our model had different values.
What if the latest poll from The New York Times/Siena College had Biden's support 3 points higher, as a poll of 1,000 people would be expected to have about 5 percent of the time?
Or what if the third-party adjustment were smaller?
Our model answers these questions by simulating thousands of different polling averages each time it runs.
That, in turn, lets us show uncertainty intervals directly on the average — to our knowledge, the first time a general-election polling average has done so since HuffPost Pollster in 2016.
Again, though, these uncertainty intervals aren't showing you the uncertainty that arises from the possibility of an industrywide polling miss.
That is the job of our forecast.
The uncertainty in our averages answers the question, "How different could support for each candidate be in the average poll today if we observed more polls with slightly different results?", while the uncertainty in our forecast answers the question, "How different could each candidate's actual support be from the polls?"
Conclusion
I should note that variations of this model have been proposed by political scientists and statisticians for a few decades.
(The following paragraph is not exhaustive, but should give you a good sense of the history of this approach.)
By my research, Robert Erikson and Christopher Wlezien were the first to model support for candidates with house effects and as a smooth function of time over the campaign.
Political scientist Simon Jackman later formalized a Bayesian time-series model of polls with house effects for Australian elections.
Statisticians Kari Lock and Andrew Gelman employed a time-series model of polls as part of a paper forecasting election outcomes with polls and other data.
Political scientist/survey statistician/pollster Drew Linzer combined aspects of these and other approaches in a Bayesian dynamic linear model of state and national polls of 2012 general election.
And I worked with Gelman and statistician Merlin Heidemanns to add additional poll-level adjustments as well as other factors to model the 2020 general election in this way.
538's new presidential general-election polling average goes one step beyond these previous approaches by modeling support for multiple candidates across geographies simultaneously and adding a few additional adjustments that have not been attempted before in the fully Bayesian context.
That's it for now!
If I've done my job well, hopefully you can see how this new approach closely resembles our original dual-purpose model of models, but in one neat package.
If you see any polls that are missing, or spot a methodological bug (these things do happen when we launch new averages, especially with models of this level of sophistication), hit us up with an email.
Footnotes
*More on this in the "How we adjust and average polls" section, but the basic reason is that our new model is not a mix of a weighted average and polynomial trendline — methods that need to be told explicitly which observations are subject to less noise — but rather a Bayesian time series model that directly takes the observation noise in the poll into account.
So we no longer need to weight on sample size, poll recency and the degree to which a poll is an outlier; the model handles that stuff for us on its own.
**As a methodological matter, we think how to best integrate pollster weights into an aggregation model is an open debate.
(In fact, one statistician told me we shouldn't weight to pollster effects at all — instead we should create complicated meta-models of polling accuracy across many different contests simultaneously and inject the estimated uncertainty and bias of each pollster directly into our model. Unfortunately, this is not computationally feasible for the type of live-updating statistical model we want to publish.)
We tried a few more complicated methods — such as adjusting the effective sample size of each poll to match how accurate we expected it to be — but they did not perform meaningfully better than more parsimonious approaches.
As a practical matter, what matters more than how you weight polls based on their quality is that you try to account for quality at all.
https://www.msn.com/en-us/news/politics ... 418a&ei=36
Re: POLITICS
The Washington Examiner
"White House Report Card: Biden drives his approval rating into a ditch"
Story by Paul Bedard, Washington Examiner
27 APRIL 2024
This week’s White House Report Card finds President Joe Biden in a tornado of trouble.
Let’s start with his polling.
Gallup on Friday averaged other polling that has shown just how unpopular the president is.
Biden has dropped to a 68-year presidential low, 38.7% approval.
That is nearly 10 points below where former President Jimmy Carter was at this point in his presidency, and we remember what happened in his reelection bid.
That dismal approval rating puts Biden exactly 10 points below the bare historical minimum he needs to secure reelection, according to Gallup.
Next there is the economy.
Not only has inflation dug in and GDP petered out, but union workers are publicly ripping Biden’s leadership on live TV.
And there were more examples of the president’s stumbles and mumbles.
First there were reports of aides being ordered to walk between the president and media cameras to hide Biden’s feeble gait.
There were his weird and wrong claims that he used to drive 18-wheelers.
And he’s back to reading aloud directions to “pause” on his teleprompter.
Both our graders saw troubles this week.
Conservative Jed Babbin graded it a “D-minus” and highlighted moves by the White House to hurt the economy.
Pollster John Zogby graded the week a “C” and said the president’s stumbles have blocked chances to take advantage of problems facing his political foe, former President Donald Trump.
Jed Babbin
Grade: D-
President Biden’s war on the American economy continued with more regulations this week designed to bring it to a halt.
And that’s not the half of it.
You might remember former President Donald Trump’s 2017 remark that there were “very fine people” on both sides of the Charlottesville, Virginia. “white supremacist” rally.
Biden said that sparked his entry into the 2020 presidential race.
But this week he said something similar when talking about anti-semitic protests at colleges.
He condemned the hate talk, but added, "I also condemn those who don’t understand what’s going on with the Palestinians…”
Biden is still terrified of the Reps. Ilhan Omar (D-MN) and Rashida Tlaib (D-MI) wing of the Democratic Party who won’t vote for him unless he bashes Israel and qualifies every defense of American values with defenses of the Palestinians and their ilk.
Biden also continued his EV push.
But his far-sighted regulatory regime is, at the same time, making it harder to produce electricity.
This week a final regulation was issued that requires coal and natural gas electricity producers to effectively cut all emissions by 2032.
For many the only solution will be to close, thus reducing the amount of electricity available.
If you want to shut down the U.S. economy, that’s one easy way to do it.
Meanwhile, inflation caused by Biden’s radical spending continues.
Old Joe celebrated Earth Day by subsidizing solar panels — more billions spent on nonsense.
Bidenomics continues to wreck our economy.
And let’s not forget Biden’s latest venture into “net neutrality” which will reduce competition on the internet and raise consumer prices.
Fortunately, there wasn’t a lot of international action this week, though the Middle East remains on edge following Israel’s response to Iran’s massive attack last week.
Finally, Old Joe claimed to have driven an 18-wheeler in his long career.
Of course, he never did.
And he managed to read out loud the instructions on his teleprompter to “pause” at one point.
His mental state continues to decline rapidly.
John Zogby
Grade: C
President Joe Biden and former President Donald Trump are tied in a two-way and tied in a five-way race.
Not 'virtually' tied, but really tied.
And they are likely to stay that way for the foreseeable future.
Biden should have been able to take advantage of Trump's legal and financial woes, gag orders and all.
But the current president got gob-smacked by pretty lousy economic news.
No sooner had Jamie Dimon, CEO of JP Morgan Chase, declared a boom economy, a new report showed the GDP grinding down to an annualized 1.6% growth rate and inflation jumping back to a 3.7%.
The president traveled to key battleground states touting federal spending that was generating jobs and winning adulation from the locals.
But his job approval is still at about 40% and he is still trailing in battleground states.
Jed Babbin is a Washington Examiner contributor and former deputy undersecretary of defense in the administration of former President George H.W. Bush. Follow him on X @jedbabbin.
John Zogby is the founder of the Zogby Survey and senior partner at John Zogby Strategies. His podcast with son and managing partner and pollster Jeremy Zogby can be heard here. Their firm polls for independent presidential candidate Robert F. Kennedy Jr. Follow him on X @ZogbyStrategies.
https://www.msn.com/en-us/news/politics ... fccc&ei=16
"White House Report Card: Biden drives his approval rating into a ditch"
Story by Paul Bedard, Washington Examiner
27 APRIL 2024
This week’s White House Report Card finds President Joe Biden in a tornado of trouble.
Let’s start with his polling.
Gallup on Friday averaged other polling that has shown just how unpopular the president is.
Biden has dropped to a 68-year presidential low, 38.7% approval.
That is nearly 10 points below where former President Jimmy Carter was at this point in his presidency, and we remember what happened in his reelection bid.
That dismal approval rating puts Biden exactly 10 points below the bare historical minimum he needs to secure reelection, according to Gallup.
Next there is the economy.
Not only has inflation dug in and GDP petered out, but union workers are publicly ripping Biden’s leadership on live TV.
And there were more examples of the president’s stumbles and mumbles.
First there were reports of aides being ordered to walk between the president and media cameras to hide Biden’s feeble gait.
There were his weird and wrong claims that he used to drive 18-wheelers.
And he’s back to reading aloud directions to “pause” on his teleprompter.
Both our graders saw troubles this week.
Conservative Jed Babbin graded it a “D-minus” and highlighted moves by the White House to hurt the economy.
Pollster John Zogby graded the week a “C” and said the president’s stumbles have blocked chances to take advantage of problems facing his political foe, former President Donald Trump.
Jed Babbin
Grade: D-
President Biden’s war on the American economy continued with more regulations this week designed to bring it to a halt.
And that’s not the half of it.
You might remember former President Donald Trump’s 2017 remark that there were “very fine people” on both sides of the Charlottesville, Virginia. “white supremacist” rally.
Biden said that sparked his entry into the 2020 presidential race.
But this week he said something similar when talking about anti-semitic protests at colleges.
He condemned the hate talk, but added, "I also condemn those who don’t understand what’s going on with the Palestinians…”
Biden is still terrified of the Reps. Ilhan Omar (D-MN) and Rashida Tlaib (D-MI) wing of the Democratic Party who won’t vote for him unless he bashes Israel and qualifies every defense of American values with defenses of the Palestinians and their ilk.
Biden also continued his EV push.
But his far-sighted regulatory regime is, at the same time, making it harder to produce electricity.
This week a final regulation was issued that requires coal and natural gas electricity producers to effectively cut all emissions by 2032.
For many the only solution will be to close, thus reducing the amount of electricity available.
If you want to shut down the U.S. economy, that’s one easy way to do it.
Meanwhile, inflation caused by Biden’s radical spending continues.
Old Joe celebrated Earth Day by subsidizing solar panels — more billions spent on nonsense.
Bidenomics continues to wreck our economy.
And let’s not forget Biden’s latest venture into “net neutrality” which will reduce competition on the internet and raise consumer prices.
Fortunately, there wasn’t a lot of international action this week, though the Middle East remains on edge following Israel’s response to Iran’s massive attack last week.
Finally, Old Joe claimed to have driven an 18-wheeler in his long career.
Of course, he never did.
And he managed to read out loud the instructions on his teleprompter to “pause” at one point.
His mental state continues to decline rapidly.
John Zogby
Grade: C
President Joe Biden and former President Donald Trump are tied in a two-way and tied in a five-way race.
Not 'virtually' tied, but really tied.
And they are likely to stay that way for the foreseeable future.
Biden should have been able to take advantage of Trump's legal and financial woes, gag orders and all.
But the current president got gob-smacked by pretty lousy economic news.
No sooner had Jamie Dimon, CEO of JP Morgan Chase, declared a boom economy, a new report showed the GDP grinding down to an annualized 1.6% growth rate and inflation jumping back to a 3.7%.
The president traveled to key battleground states touting federal spending that was generating jobs and winning adulation from the locals.
But his job approval is still at about 40% and he is still trailing in battleground states.
Jed Babbin is a Washington Examiner contributor and former deputy undersecretary of defense in the administration of former President George H.W. Bush. Follow him on X @jedbabbin.
John Zogby is the founder of the Zogby Survey and senior partner at John Zogby Strategies. His podcast with son and managing partner and pollster Jeremy Zogby can be heard here. Their firm polls for independent presidential candidate Robert F. Kennedy Jr. Follow him on X @ZogbyStrategies.
https://www.msn.com/en-us/news/politics ... fccc&ei=16