One speculative source of the political change is the defunding of U.S. Agency for International Development (USAID) in 2025. The implication is that USAID’s philanthropic grantmaking towards civil society groups and nation building is just soft power political influence in disguise meant to prop up favored candidates and oust unfavored ones akin to the color revolutions in Eastern Europe.

Yet there’s been little evidence found of USAID or other grants related to election interference. One potential State Department grant in 2026 related to countering censorship got a particular amount of attention and complaints from Brazil’s attorney general, but that was only for $1 million—a small amount considering the hundreds of millions sometimes handed out on a yearly basis—with no recipient being designated yet.

President Trump imposed sanctions on Brazil over prosecution of ex-president Bolsonaro, which led to accusations of electoral interference, but it didn’t seem to affect the outcome of the prosecution.

And much of USAID funding in Brazil doesn’t appear to be aimed at civil society groups or media campaigns in general. USAID did heavily support the Consortium for Elections & Political Process (CEPPS)—a nonprofit focused on strengthening democracies and judiciaries around the world—with $3.3 billion dollars in grants, but none of those grants were directed towards Brazil based on USASpending data.

Instead USAID funding is largely directed at the Amazon, related to subjects like environmental preservation, indigenous outreach, and support for Venezuelan refugees. Those might seem like benign philanthropic aims from a distance, but they are highly controversial topics in Brazil. Bolsonaro’s base surrounds the industrial and agribusiness center of Brazil around Manaus, whereas da Silva has regularly courted preservation groups.

Much of the refugee funding goes through the Adventist Development and Relief Agency International, who has received over $40 million since 2008, largely related to refugee support. The Task Force for Global Health has received about $39 million for vaccines and medicines to help eradicate tropical diseases like river blindness. The Instituto Internacional de Educação do Brasil (IEB) is a civil society training individuals in environmental concerns and indigenous cultures that received over $33 million. The U.S. environmental organization The Nature Conservancy received over $13 million. The Carter Center received almost $21 million to focus on river blindness, voter education, and educating women.

Funding detailed here is for assistance and does not include contracting payments. Amounts are based on obligations not necessarily distributed yet and can include loans along with direct grants.

Battle Over Belo Monte Dam

In general, USAID did little in Brazil prior to 2016. In 2016 USAID grants flooded in, with a $9 million grant to Global Initiatives for its Prospera program that funded sustainable corn production as well as initiatives to “combat hunger, promote gender equality, and reduce inequities in the communities surrounding Coromandel and São José do Belmonte.”

That was under then-president Dilma Rousseff—the ex-Marxist guerilla with the Armed Revolutionary Vanguard-Palmares (VAR-Palmares) who went on to work in the energy department under the first da Silva cabinet from 2002 to 2010, before running for president herself in 2010.

Large U.S. funding to the Amazon during her tenure is interesting as her presidency was embattled with indigenous groups over development of the 10 gigawatt Belo Monte mega dam on the Xingu river in Pará—a tributary of the Amazon. Belo Monte was conceived back in the 1970s but struggled against environmental opposition for decades, and it wasn’t until Rousseff championed the project during her tenure that any headway was made, giving it approval in 2011 soon after election with operation beginning in 2016. It now produces somewhere between five and 10 percent of the country’s total energy generation depending on the time of year.

Rousseff would be dogged by accusations of bribery related to the project for years, and indigenous groups continue to seek reparations. She was eventually impeached in 2016 over accusations of falsifying the government’s finances to look better than they actually were. While Rousseff was at odds with indigenous groups over environmental concerns, she is close to da Silva who actively supports Amazon anti-deforestation campaigns.

That time period under Rousseff is also when the Brazil security agency, Agência Brasileira de Inteligência (ABIN), wrote reports in 2011 and 2012 detailing U.S. nonprofits interfering in Brazilian affairs by supporting indigenous campaigns against infrastructure projects in the Amazon, like the Belo Monte dam, the BR-317 portion of the Interoceanic highway in Acre, the Urucu-Porto-Velho gas pipeline, and hydroelectric plants on the Madeira river over social and environmental concerns. The reports were not revealed until 2023 through a Senate inquiry.

It called out the World Wildlife Fund (WWF) and Nature Conservancy—two USAID Brazil grant recipients—along with various U.S. and Brazilian nonprofits like Amazon Watch, Greenpeace, Avaaz Foundation, Amigos Da Terra (Friends of the Earth Brazil), International Rivers, Clinton Global Initiative, Friends of the Earth, and Conservatism International for their work campaigning against the Belo Monte dam project.

Much of USAID’s funding during Rousseff’s period had died down by the time the dam was completed and Jair Bolsonaro took office in 2019, only to see over $120 million spent in Brazil around the same time as the reelection of Lula in 2023.

State Department Funding Eclipses USAID

Whether or not USAID was influencing Brazilian elections seems irrelevant since much of USAID has been integrated into the State Department, which now spends significantly more in Brazil than what USAID ever did.

USAID funding to Brazil topped off at $62 million in 2024 while the State Department now spends over $125 million a year there, and that doesn’t include funding from various other agencies such as the Department of Health and Human Services, Overseas Private Investment Corporation, Department of Agriculture, National Science Foundation, and the Department of Defense among others.

Funding Labor More Than Environmental

Historically the two agencies funded similar topics in similar amounts, like environmental and refugee causes. USAID tended to fund more related to healthcare whereas the State Department put more towards education and human rights.

But in the last two years, State Department funding has gone more to labor organizations in Brazil than environmental ones. Top recent recipients include:

• U.N.’s International Labor Organization ($39 million)

• Vital Voices Global Partnership ($22 million) to support survivors of gender-based violence

• Fundação Para O Desenvolvimento Científico E Tecnológico Em Saúde (Fiotec) for HIV Support ($20 million)

• United Nations Development Program ($13 million) for climate observations

• American Center for Labor Solidarity, connected to the AFL/CIO ($12 million) to support labor organizations

COVID Funding Back and Forth

The largest funding on record is from the U.S. International Development Finance Corporation in 2021 of $250 million, but that was for government risk insurance related to purchase of COVID vaccines. While that was valued at $250 million, in effect Brazil probably gave more money to the U.S. through the purchase of COVID vaccines, with estimates in the range of $21 to $28 billion.

Bond yields have been intertwined with rising oil prices, and because bond yields are inversely correlated with bond prices, prices for those bonds have collapsed, leading to a collapse in all other bond markets, not just Treasuries.

But that correlation is new. Prior to 2020, oil prices went up and bond yields went down. Something switched in 2020, and that mystery element is the Federal Reserve’s interest rate.

The Federal funds rate is one of the handful of levers the Federal Reserve wields to maintain the economy, like the discount rate, to hinder or unleash inflation. It sets the interest amount that the Fed pays on money lent out to other lending institutions.

And the Fed has been raising interest rates since 2020 in sync with the rising yield and falling price on bonds. Prior to that, interest rates trended downward alongside Treasury yields for decades since the Volcker era of the early eighties. Prior to 1982, they both trended upward.

Yields and interest rates aren’t always exactly in sync. Like after the financial crisis when interest rates were effectively zero and bond yields were in steady decline. But they trend together, and they zig and zag at times of sudden interest rate changes.

Low Rates Equate To Other Lower Rates

The connection between Treasury yields and interest rates isn’t direct but simply a remnant of how interest rates drive many aspects of the economy.

Federal Reserve interest rates set the amount that investors earn on their deposits at Federal Reserve banks. Those same investors could also be investing in Treasury bonds, and essentially the two products are in competition. If the Treasury needs investors in their bonds, the yields will have to be competitive with the rate they would get by simply parking their money in a Federal Reserve bank overnight—the easiest option for a reliable return.

When bonds are said to follow inflation, there is some truth to that too. Interest rates are often raised to counteract inflation, which would also raise bond yields—potentially what is happening now. But there is no guarantee as it is up to the whims of the head of the Federal Reserve whether or not to raise interest rates.

Whether or not bond prices collapse or not is almost moot considering that it may not be a bad thing for current investors. Low bond prices mean that it costs less to buy a bond that pays out a higher percentage. But it certainly affects current temporary holders of Treasuries trying to sell them at current rates in the secondary market.

For example, China buys lots of Treasuries with money received from trade exports, not for their yield, but as a way to keep their money in dollars, depress the value of the yuan relative to the dollar, and encourage more exports—sometimes referred to as the Export-Treasury Feedback Loop.

In a recent interview, U.S. Treasury Secretary Scott Bessent stated that to potentially alleviate economic concerns interest rates will likely come down once the U.S.-Iran conflict over the Strait of Hormuz is over.

The conflict there has driven up oil prices over the last few months, which in turn has driven up the yield on Treasury bonds in sync.

The U.S. has had some of the highest correlations between Treasury rates and crude oil prices as the two prices have been intertwined. The monthly average price of West Texas Intermediate crude tracked the yield on the 10-year Treasury bond almost exactly, with a Pearson correlation of .9293.

The Treasury recently increased interest rates to combat the potential resulting inflation. The general assumption is that bond yields follow crude oil prices because they are an indicator of inflation. High oil and gas prices lead to higher costs of everything because so much of the economy relies on oil and gasoline products.

That, in turn, leads to cheaper value of Treasury bonds because inflation erodes the value of the bonds. Cheaper bonds lead to higher yields as the two values are inversely correlated.

But oil prices are not commonly an indicator of inflation. In general, oil prices rise with inflation over the course of a decade like any commodity does, but the connection is loose.

There may be times when they are in step. For example, during the pandemic oil prices crashed and then swung upward severely, and so did inflation. But there are plenty of times when they are not even close.

Previously Little Correlation Between Oil, Bonds, And Inflation

At other times oil prices can fluctuate wildly over the course of a year with no real substantial effect on inflation. Prior to the pandemic, inflation was relatively consistent for decades despite all kinds of oil price fluctuations.

For example, crude oil prices collapsed beginning in 2014, falling from over $105 per barrel to almost $30. Interest rates remained minimal throughout that time and inflation for 2014 through 2017 was about the same—1 to 2 percent. The one exception being 2015 where inflation was a low .1 percent.

In the early 2000s, energy prices grew significantly, doubling between 1999 and 2006. But inflation over that period was relatively consistent. In the early nineties, energy prices remained low while inflation was relatively consistent there too.

Inversion of Bond Yields

The correlation between Treasury yields and energy prices or inflation in general is something relatively new. Going back to the 1980s, 10-year Treasury bond yields were in steady decline despite oil prices generally rising across most of that time period. It’s only since around 2020 when that metric inverted.

Essentially, for forty years prior to the pandemic the average inflation was low. Treasury bonds were used more like stocks whose value is in their rising market price rather than as a long term investment based on interest payments until maturity.

If anything, the new regime of bond pricing is a reversion to what it was in the 1970s prior to the Volcker takeover of the Federal Reserve. Back then, oil shocks not only sent gasoline prices sky high, but also inflation and Treasury yields in tandem.

Why that happens now and in the 1970s but not in the years between is the subject of numerous economic papers with few concrete conclusions.

Alt-A, or Alternative A-paper, loans in particular were often described as “liar loans” where the borrowers misled the lenders about how much they earned and were almost as prolific as subprime loans.

Sometimes they are also alt-doc, or alternative documentation, loans, where the borrower provides alternate proof of their ability to pay back the loan. For example, a number of home equity lines of credit (HELOC) loans might be considered alt-doc loans as the borrower doesn’t use proof of income but just proof of home ownership to borrow against.

Alt-A are often jumbo loans with large loan-to-value (LTV) ratios, say for someone borrowing enough to for their current mortgage and a second house at the same time. A loan using their current home equity as collateral could be both Atl-A and Alt-doc.

Alt-A and Alt-doc lending is reasonable except that much of the lending surrounding them and subprime loans was fraudulent during the subprime crisis, not because they used alternative information as collateral but because the applicant lied about everything in their paperwork.

According to a 2008 report from the New York Federal Reserve, real-estate fraud investigations doubled between 2001 and 2003, and suspicious activity reports (SARs) filed by federally-regulated institutions related to mortgage fraud increased from 3,500 in 2000 to 28,000 in 2006.

The Mortgage Asset Research Institute (2007) estimates that direct losses from mortgage fraud exceeded $1 billion in 2006, more than double the amount from 2005.

For example, subprime and Alt-A loans originated in 2006 have experienced historical levels of serious early payment default (EPD), defined as being 90 days delinquent only three months after origination.

It wasn’t simply falsifying income on a loan application but a wide range of misrepresentation, coordination, and various other scams and ruses being perpetrated.

False appraisals, fake down payments, ghost borrowers and air loans where the lender invents borrowers out of nothing, or buying a home and renting it with no intention of making mortgage payments. Brokers took out loans in potential investors’ names without the investors knowing that they signed up for a loan. Borrowers would work with lenders to obscure the lack of a down payment and hide how little capital the borrower had on hand.

So much of the lending in the private markets was based solely on the borrower’s self-reporting; not just income, but other determiners of creditworthiness like owner-occupancy and borrower debt.

Owner-occupied properties in particular might be very drastically misrepresented. According to a 2009 paper in the Journal of Economic Perspectives, investor-owned or non-owner-occupied properties only accounted for about 8 percent of subprime loans.

But based on the top three centers of the subprime crisis, it could be much higher than that. The number of vacant properties added each year in Lee, Clark, and Riverside counties was regularly more than the owner-occupied additions, making it seem like the rate of investor-owned properties added was more on the order of 50 percent than 8 percent.

Misrepresentation of owner-occupancy is not a new concept as a 2013 paper from NBER noted that 6 percent of mortgage loans for owner-occupied properties were actually investment properties. Other investors lied about having secondary liens in their name.

More than 6% of mortgage loans reported for owner-occupied properties were given to borrowers with a different primary residence, while more than 7% of loans (13.6% of loans using a broader definition) stating that a junior lien is not present actually had such a second lien. Alternatively put, more than 27% of loans obtained by non-owner occupants misreported their true purpose and more than 15% of loans with closed-end second liens incorrectly reported no presence of such liens.

While documentaries on the financial collapse often talk about the mortgage-backed securities that collect loans into piles which get investment grades and then sell off tranches which are then used for CDOs and swaps. Supposedly, all the risk was buried and intentionally misrepresented because so much money was being made. Ratings agencies were simply rubber-stamping positive AAA ratings because they didn’t want to lose business to other ratings agencies.

That seems hard to believe as significantly falsifying the grade of an investment on a large-scale would be massive fraud on the part of the ratings agencies; not just altering a few loans but thousands of loans. The ratings agencies would make settlements in the millions in related cases, but no real evidence appeared that they manipulated their numbers.

The model that Moody’s used for valuing their subprime loans included all kinds of potential flags that might highlight a risky loan in a tenuous housing market, like debt-to-income ratio, local housing market projections, adjustable interest rate, interest only and negatively amortizing loans, owner occupancy, mortgage insurance, coupon/interest rate, documentation status, and lien type.

But all those metrics would be meaningless if the underlying data was inaccurate. For subprime lending, a lot of it was self-reported. Compare that to Federal Housing Administration (FHA) mortgages that get underwritten by Fannie Mae and Freddie Mac—e.g. not subprime—where income is verified by IRS filings and other government documentation.

Private loan insurers underwrote many of the subprime loans, but they likely did so based on the documentation submitted. Loan originators like Countrywide were blamed for being the first point of entry for subprime loans, who then sold off the loans into the larger market. But Countrywide would have been dependent on the mortgage brokers to validate documents and information in preparation for a loan.

A 2011 Miami Herald story detailed how thousands of felons were signing up to be loan originators in the state because of lax rules for the profession. Another Herald story from 2008 details how the state’s Office of Financial Regulation looked the other way for mortgage brokers’ licensing requirements, skipping background checks, and ignoring signs of crooked mortgage operations. There were repeated attempts to have loan originators licensed, but the industry pushed back.

In 2008, Florida Governor Charlie Crist would block felons from becoming mortgage brokers and push the chief of the state’s Office of Financial Regulation to resign.

California and Nevada weren’t as bad, but they had similar issues. There were no license requirements on brokers in California, and Nevada was rife with mortgage scams and straw buyers.

In the Las Vegas Mazzarella-Grimm case, a couple used a network of LLCs to buy up $107 million worth of houses using straw buyers and sell them to each other to push up home prices.

Since the foreclosure crisis, numerous statues have been put in place to prevent predatory borrowing like this, like the 2008 Secure and Fair Enforcement for Mortgage Licensing Act (SAFE Act) requiring licensing of mortgage brokers and originators, background checks, and validation of owner-occupancy. Dodd-Frank required lenders document a borrower’s ability to pay through the ATR/QM Rule.

Florida would enable fast-tracking of foreclosures, particularly vacant ones by throwing money at the court system, while California and Nevada already had non-judicial foreclosure rules in place.

Homes became overvalued. As previously covered in Investigative Economics, when interest rates went up, borrowers could no longer afford their monthly payments. The bubble burst, and the subprime crisis led to the foreclosure crisis. The wave of foreclosures in sunbelt cities left thousands of vacant properties in its wake as homeowners found homes somewhere cheaper.

But the vacant property rate was actually higher before the foreclosure crisis for some of the municipalities hit hardest by the foreclosure crisis, not after, and it was growing fast. On top of that, owner-occupied housing grew little in a period that was supposedly driven by home ownership. For some areas, the owner-occupied housing rate actually declined based on Census American Community Survey (ACS) estimates.

Ostensibly, all those subprime borrowers were living in the homes they were buying, but the numbers seem to indicate they were being bought but not lived in as investments.

Sunbelt Housing Boom

For example, the top three regions hit the hardest by the foreclosure crisis are the following:

• Cape Coral-Fort Myers, Lee County, Florida—on the West coast of Florida, south of Tampa—where more than 40,000 foreclosures happened in 2008 alone.

• Las Vegas, Clark County, Nevada—over 40,000 foreclosures between 2007 and 2008.

• Inland Empire, Riverside County, California—Southeastern California, East of Bakersfield and close to Nevada—69,855 foreclosures during the crisis.

But Census housing data from the respective three counties shows the owner-occupied rate was in decline back in 2006 before the wave of foreclosures hit. Instead of growth in owner-occupied housing, vacancies were on the rise.

Eventually there was a noticeable population exodus from these counties, potentially as prospective homeowners looked elsewhere, but it began in 2010.

Construction Led The Housing Bubble

Likely, much of that is simply the result of the new housing construction boom. Housing permits exploded during the bubble, likely taking advantage of low interest rates at the time, and sometimes it takes time for a newly constructed home to be bought and occupied.

Some were likely rented out, as total rental units grew over the time period, more so than owner-occupied housing.

Subprime Borrowers Were Investors and Flippers

Many of these are unsurprisingly subprime loans, which often gets equated with high-interest lending, but subprime lending also just means any loan that doesn’t get securitized by Fannie Mae and Freddie Mac for other reasons, like a lack of documentation or a large-amount, short-term loan that balloons quickly. Which is why the collapse of mortgage-backed securities (MBS) only happened in the private or shadow MBS market and didn’t really touch the financial markets that major pension funds and index funds invest in.

If subprime borrowers were simply a large population of low credit Americans hoping to own a homestead to call their own, the Troubled Asset Relief Program (TARP) was supposed to lend them a hand and keep a roof over their heads.

But TARP didn’t seem to target anybody like that. The largest recipient state was North Carolina—not exactly an epicenter of the subprime crisis—and few recipients listed factors like high interest rates commonly associated with subprime lending as why they needed help. Most simply stated “curtailment of income” as the cause.

So, who were the subprime borrowers then? They bought mainly new houses without mortgages for homes they didn’t live in. When foreclosed upon, they didn’t seek subsidies in any significant numbers.

A large share were investors. According to a 2018 paper from the National Bureau of Economic Research (NBER), real estate investors had their share of mortgage balances double between 2004 and 2007, who then accounted for a large portion of foreclosures that occurred before 2010 and before the significant growth in unemployment:

Real estate investors have higher default rates than regular borrowers and they accounted for close to 50% of all foreclosures at the height of the crisis, even though their share in the borrower population peaked at 14%.

The Investor Speculator Bubble

While the crux of the financial crisis is still the same—that banks handed out subprime loans to those less likely to pay for an unsustainable price on houses—the subprime housing bubble was a remnant of investor speculation and not owner-occupiers looking for a slice of the American dream that defined the TARP bailout.

Movies like The Big Short portrayed the bubble as one of Wall Street wrapping up shoddy mortgages that fed into the major financial markets, poisoning the well of investment and requiring a federal bailout to prevent a catastrophe. But the majority of subprime MBSs lived in the private, shadow securitization market that was separated out from the major markets of Wall Street. Fannie Mae has started dipping their toe into subprime loans right at the beginning of the collapse, but it wasn’t much.

Instead, maybe the best description of the situation is this 2009 New Yorker story, The Ponzi State, which details how Florida is rife with speculative housing cycles not unlike every other speculative financial bubble from tulips to beanie babies.

When interest rates were low, retirees flooded into the state and everybody became a broker, buying up houses on weak credit to flip for a higher price. Tens of thousands of dollars can be made overnight by anybody willing to take out a loan and borrow against personal equity to take a chance on a house they will never live in.

Prices go up and up unless interest rates go up, as they did beginning in 2003, causing short-term flippers to make substantial interest payments, or when pensioners stop moving to Florida.

A very similar situation happened in Florida in the 1920s when swampland salesmen hyped up homes that helped feed into the financial collapse of 1929, leaving semi-empty housing tracts behind.

But at the same time, the fear over subprime lending also helped foment a bank run in places where there wasn’t overly speculative housing construction built on subprime lending.

Since 2022, total energy generation has increased 128 terawatt-hours for the first six months of the year (January to June), or about 6 percent. In July, regional transmission operator (RTO) Pennsylvania, New Jersey, and Maryland Interconnection (PJM) recorded some of the highest auction prices for electricity in its history. For some households, that led to a stark increase in electricity costs.

Soon after, Pennsylvania Governor Josh Shapiro sued PJM to cap their auction prices and prevent rate hikes on customers. PJM pointed to the ongoing supply-demand imbalance as the source of high auction prices.

While the increase in demand in AI is substantial, net capacity—total potential energy generation—has been increasing year over year. According to Energy Information Administration (EIA) data, about 69 gigawatts of capacity was added between 2022 and 2024, with a potential additional generation of 604 terawatt-hours per year—more than enough to cover the growth from AI.

But capacity is sometimes a loose term. It can mean what a coal plant could theoretically deliver if it was run constantly for 24-hours, or it can mean what a solar panel can deliver at peak sunlight.

Those situations are not always possible. Coal plants go offline for repairs. Solar panels can never obtain full sunlight for 24 hours at a time.

For example, China often boasts of its installed solar capacity that easily dwarfs that of its coal generation capacity. But actual generation numbers show that China still gets much more of its energy from coal. China’s massive solar capacity doesn’t equate to massive solar generation. That’s because China’s capacity factor for solar—the ratio of what it actually generates to its theoretical nameplate capacity—can be as low as ten percent. In the U.S., average capacity factor for solar hovers around 20 to 30 percent, depending on where it is installed.

That’s why PJM and other RTOs use an adjusted metric for what is considered actually available capacity, such as unforced capacity (UCAP), which is equal to:

Where EFORd is the Equivalent Demand Forced Outage Rate—the pre-calculated rate at which an energy source is unavailable.

In PJM’s reports they have two measurements of unforced capacity: UCAP, which refers to the adjusted capacity available for market auctions, and Fixed Resource Requirement (FRR)—generation capacity at a fixed price for certain utilities with PJM that choose to not use the auctions for obtaining electricity.

In PJM’s capacity auctions, generators bid to provide capacity—energy that may or may not get used at a time in the future. If the energy is not needed, the generator still gets paid. But if the energy is needed and the generator can’t provide it, they can be penalized through deficiency charges.

In 2022, Dominion Energy of Virginia chose to leave the PJM markets and switch to FRR for their customers. Dominion is also the utility that provides energy generation for Loudon County where the largest collection of data centers resides. But in 2025 they returned to the UCAP markets.

ELCC Changes

In total, PJM’s capacity wasn’t really affected by Dominion leaving the auction and returning. Instead the drop in capacity in 2025 was largely a result of Federal Energy Regulatory Commission’s (FERC) marginal Effective Load Carrying Capability (ELCC) rule in 2024.

In 2022, Winter Storm Elliott almost led to rolling blackouts across the rust belt and exposed the lack of true capacity in energy markets. As a result, FERC reinterpreted how it valued capacity in energy markets to better weigh tight or marginal situations. The final marginal ELCC rule would weigh more resilient forms of energy over others, and some generators would bow out of capacity auctions as a result.

As part of ELCC rules, each generation source gets an ELCC rating that is effectively like a capacity factor—how reliable the source is to generate energy when needed—and the new rules set the ELCC ratings for solar much lower when they were.

With the new changes, tracking solar—solar panels that track the sun’s movements—would have its ELCC rating fall from 54 percent to 14 percent. Fixed-tilt solar would go from 36 percent to 9 percent. Battery storage, which was previously given ratings around 100 percent, would lose 20 to 30 percentage points. Wind would actually have its ratings upgraded for both onshore and offshore turbines.

The lower rates pushed the accredited UCAP much lower and make bidding in the capacity auction less appealing, with somewhere on the order of 34 gigawatts disappearing from the market as a result. The remaining generators can then charge more for the dwindling supply

It is constantly the target of shareholder activism over investment decisions that not only affect the retirement benefits of thousands of state workers, but they set the direction of pension plan investment nationwide.

A recent report funded by the Retired Public Employee's Association of California (RPEA) showed the plan chronically underperforms compared to its peers, especially considering its emphasis on alternate investments like private equity with little disclosure on fees. Other times, it is embattled over its Environmental, Social, and Governance (ESG) investments, and whether those are necessary to save the planet or simply throwing money down the well.

But little gets mentioned about how much CalPERS, and potentially other pension plans, are now paying towards health benefits since the passage of Obamacare.

What was once less than $2 billion a year in healthcare claim costs and administrative expenses for CalPERS before 2013 has doubled to become $4 billion and is now over $5 billion a year.

Some of that is related to flex-funding, where the pension fund pays for capitation—a specific set of services like primary care and some fee-for-service healthcare—up to a specified amount. Private insurers handle the rest. While claims spending has gone up, so have the premiums collected by CalPERS to pay for them, and the net balance has not changed all that much.

According to the CalPERS annual financial report, the 39.3 percent growth in claims expenses in fiscal year 2014 to 2015 was largely driven by flex-funded health plan activity and the 81 percent growth in administrative costs came from Affordable Care Act (ACA) tax payments.

Flex-funding was meant to help drive down healthcare costs. Rather than leaning on insurers to keep costs down, the pension fund handles claims and can see and control what claims get approved and for how much. They get to keep any net spending that doesn’t otherwise get used rather than the insurer.

But independent of flex-funding, CalPERS is still throwing billions at private insurers—mainly Health Maintenance Organization (HMO) plans like Kaiser Permanente—and Medicare buy-ins for the other end of healthcare for its members, upwards of $10 billion a year.

In terms of net healthcare spending, between Medicare, HMOs, and flex-fund spending and premiums, CalPERS is likely spending on order of $6 billion more per year in inflation-adjusted dollars than they were in 2008.

It’s a decent chunk of change considering that the pension fund earned $61.4 billion in investment returns in fiscal year 2024-2025.

Not Just More Enrollees

Even when accounting for the growing number of CalPERS healthcare plan enrollees, the spending is growing off the charts. What was once on the order of $6,500 per enrollee on private plans is now about $9,000 in 2024 dollars.

Medicare buy-in spending per enrollee has actually declined in inflation-adjusted dollars, although those plans benefit from government subsidization of Medicare.

While $9,000 for annual insurance might seem cheap considering that family plans can run into the tens of thousands of dollars, that is an average across all enrollees, individuals and families. Plus, large pension funds should pay lower rates on average as they leverage their large populations for discount rates.

Phrases like “Mike’s Nature trick” were references to ongoing issues in years after 1961 where tree ring data no longer correlated with atmospheric temperature data—otherwise known as the “divergence.”

“I’ve just completed Mike’s Nature trick of adding in the real temps to each series for the last 20 years (ie from 1981 onwards) and from 1961 for Keith’s to hide the decline.”

Critics pounced on the language which seemed to implicate some kind of coordinated fraud or unethical massaging of the data.

But scrutiny of the research following Climategate, including an investigation by the National Science Foundation (NSF), would eventually absolve the researchers of any accusations of manipulation.

And the divergence issue that was ostensibly being manipulated was largely irrelevant. While divergence is a well-established issue for some tree ring data, not all tree ring data sets have the problem, and it is assumed to be a relic of local conditions for specific tree populations, like droughts and other various theories.

North Cairngorms’ Tree Ring Example

For other tree ring data sets, like the one from North Cairngorms, Scotland (NCAIRN), tree ring data tracks quite well with historic temperature measurements like that of the Central England Temperature (CET) set, which goes back to 1659 AD. Divergence is not an issue there.

The decennial average of atmospheric temperature from CET correlates significantly with the decennial average of temperature anomalies based on the Scottish tree ring widths (Pearson: .8064).

That aligns with the research from climatologist Keith Briffa—who was on the Climategate email exchange—showing the sensitivity of tree growth to temperature. His data showed even higher correlations of over .90 at times depending on the tree ring dataset.

But what was left unsaid is that, if tree ring data is a valid estimate of temperature, then the current day warm period is not that different from prior warm periods, like the Medieval Warm Period around 1300 AD—only off by less than one degree celsius.

It would also be similar to the other warm period in the late 1500s that doesn’t have a unique name.

Tree Ring Data Vs. Mann’s Hockey Stick Graph

What might be more controversial is that while tree ring data tracks well with historic temperatures, Michael Mann used his own model that incorporated not just tree ring data, but ice core data, corals, lake sediments, historical documents, and other records to develop his “hockey stick” chart showing a much more significant and anomalous uptick in recent temperature compared to prior centuries. In that model, prior warm periods, like the one in the 1500s, don’t appear at all.

While tree ring data is relatively well established as a reliable source of temperature data, minus divergences, ice core data has its own issues when used on more recent time frames.

Cuccinelli vs. Mann’s Source Code

As impactful as Mann’s research is on climate research at-large, his methods for incorporating numerous temperature sources are unknown.

In 2010, then-Virginia attorney general Ken Cuccinelli demanded a wide range of documents from Mann because of the implication of fraud from the Climategate emails.

Cuccinelli used a Virginia anti-fraud statute to demand all communication used by Mann to apply for taxpayer funded grants at the University of Virginia as well as all source code and algorithms used to generate the research.

Eventually the Virginia Supreme Court ruled against Cuccinelli that the state had no right to Mann’s files and the algorithm was never disclosed.

There is little data on the population of mentally ill in prison, and the data that is there is inconsistent in its definition of mental illness. The larger majority of psychiatric institutions were elderly and many were women—two populations not commonly associated with crime and prison. Local jails could certainly have shouldered the burden of the untreated mentally ill, but there was little systematic evidence to measure and confirm it.

Instead, much more data points to deinstitutionalization affecting the sharp growth in homelessness that began in the 1980s. While data there is also spotty—there were few surveys of the homeless population back then and even fewer of mental health issues amongst the homeless—there are plenty of anecdotal descriptions showing the connection.

A 1985 Government Accountability Office (GAO) paper noted that they had no accurate estimate of how large the homeless population potentially was at the time—either between “250,000 and 350,000 by the Department of Housing and Urban Development (HUD) to a high of 2 to 3 million by the Community for Creative Non-Violence, a Washington, D.C.-based advocacy group”—or how fast it was growing—somewhere between 10 and 38 percent per year.

But the consensus was that it was growing relatively fast and was visibly noticeable in major cities. The top two major causes were growing unemployment in the late 1970s and early 1980s and deinstitutionalization.

The paper also admits gauging the mental health of the homeless population is difficult, but the two are commonly connected, noting a National Institute of Mental Health (NIMH) survey showing 50 percent of the homeless having severe mental illness as well as high percentages of drug and alcohol abuse. Other common estimates put it at one-third of the homeless having mental health issues.

But deinstitutionalization started in the 1950s and was effectively complete by the mid-1970s following the Medicaid Institution for Mental Diseases (IMD) exclusion in 1965. The IMD exclusion prevented Medicaid from funding inpatient mental health and substance use treatment in residential facilities or psychiatric hospitals that have more than 16 beds, specifically for adults aged 21 through 64.

There were still a few mental institutions being emptied into the 1980s, but before the 1980s homeless crisis. And again, the majority were elderly or female; Not a common population for the homeless or imprisoned with mental health problems, many of whom were younger men with schizophrenia or substance abuse problems.

Instead, it was the economic recessions of the late 1970s and early 1980s that, combined with deinstitutionalization, drove homelessness in the early 1980s. Major psychiatric institutions were no longer there to help shoulder the burden of those suffering from mental health issues as a result of joblessness, homelessness, or any other personal crises.

A 1984 American Psychiatric Association (APA) study showed that “a substantial number of men who in years past would have been long-term residents of state mental hospitals but now have no place to live.” The few community-based care facilities that was ostensibly supposed to replace major psychiatric institutions was ill prepared to handle them, and state hospitals tightened their admittance policies.

Potentially the best description of what happened comes from the book Over the Edge: The Growth of Homelessness in the 1980s. During the 1981 to 1982 recession, soup kitchens and emergency shelters saw the large influx of homeless. Based on estimates from Housing and Urban Development (HUD) in 1984 and one from Burt and Cohen in 1987, homelessness doubled in the span of three years.

Previously, homelessness was largely confined to those who couldn’t afford a roof over their heads and lived in subsidized housing. But the 1980s saw the growth of actual street living. The causes were often a complex portrait of personal strife, financial struggles, mental illness, and drug and alcohol abuse.

The economic downturn with less affordable housing and fewer options for psychiatric treatment or psychiatric stays meant that those on the edge of homelessness and mental health crises were pushed over the edge and wound up on the street.

A 2019 Stanford paper showed little discernible effect:

We find that the voter ID law did deter those without identification from voting. And we find this effect is small enough that studies of aggregate turnout would be unable to detect it.

A 2015 Government Accountability Office (GAO) report found some decline in a few states, but it wasn’t much—at most 3.2 percentage points:

GAO found that turnout among eligible and registered voters declined more in Kansas and Tennessee than it declined in comparison states—by an estimated 1.9 to 2.2 percentage points more in Kansas and 2.2 to 3.2 percentage points more in Tennessee

A meta-analysis by the Massachusetts Institute of Technology (MIT) found that the effect of voter ID laws on turnout as “unclear.” A University of California at Davis paper found only “modest turnout effects and only minor differences across politically relevant groups.”

None of those conclusions align with the ongoing and contentious political debate over voter identification. Depending on which side you are on, either millions are illegally voting or millions of legitimate voters will be disenfranchised by the requirement of presenting an ID.

Somehow a change with negligible effects has been the heart of numerous lawsuits from groups like the American Civil Liberties Union (ACLU) and Supreme Court decisions at both the state and national level, and it currently surrounds the debate on the SAVE America Act.

Following a recent speech by President Trump on voting integrity, the Department of Homeland Security released a memo showing that “400,000 deceased registrants and over 28,000 non-citizens who illegally registered to vote”—two examples of populations that would likely be prevented from voting with an ID restriction.

A number of states passed voter ID laws around the same time in 2011, and many of them saw a drop in eligible voter population (VEP) turnout—the number of ballots cast over the number of eligible citizens in that state—for the 2012 general election based on data from the Statistics of the Congressional Election and the University of Florida’s Election Lab.

But so did states like California , Maryland, and Illinois, with nothing close to a voter ID law on the books. Essentially turnout was low across the board in 2012, and having a voter ID law didn’t make much of a difference on top of that. Midterm elections saw a similar trend although it’s complicated by infrequent gubernatorial elections that can skew turnout.

Regressing to the Mean

If that seems odd, also consider that for many of the same states that passed voter ID laws, their turnout rates suddenly fell in line with the national average in a way that seems statistically unlikely.

In prior years, turnout percentage could vary wildly from state to state and election to election. In 1980, it could be in the low 40s for some southern states like South Carolina and Georgia, or it could be over 71 percent in Minnesota. Between the 1988 election and 1992 election, Alaska’s turnout jumped 10 percentage points.

But many states are now almost in exact alignment with the national average, particularly since 2008. States like Alaska, Georgia, Idaho, Illinois, Louisiana, Rhode Island, South Dakota, and Wyoming are now repeatedly within less than 1 percentage point of the national average.

It seems reasonable that voter turnout might become more consistent from state to state over time, but a sudden transformation in one year to being in sync with a national average with such precision appears incredibly anomalous.

While the states that closely follow the national average happen to have some version of voter ID, some states without voter ID aligned pretty closely as well, like Pennsylvania and Washington, D.C. For D.C., the 2016 turnout rate was within .1 percentage point of the national average. Compare that to 2004 where it was off by over 6 percentage points from the national average.

California’s Counterexample

The trend is not in every state as many are like California, where turnout has not regressed to the mean.

California also provides more evidence that other factors may be skewing turnout more than voter ID laws since the Golden State has had a significantly lower rates than the national average in recent years—by 4 to 5 percentage points, except for 2020—despite having no voter ID laws on the books.

Diminishing Variation

Over the course of a few decades, the standard deviation of turnout rate for those select states has collapsed. Not only are they more in line with the national average, but they are also not varying from election to election anymore.

Ascribing the trend to voter ID seems fraught because the trend doesn’t make much sense. Why a select number of states would have their turnout rates stop varying while others stayed the same is an oddity.

In February, The Washington Post conceded the beepocalypse—the largescale collapse of honeybee populations related to colony collapse disorder akin to “bee AIDS”—was all hype.

This lines up with what Investigative Economics published in April of last year that showed no real collapse in honeybee populations except for those following the decline in subsidies for honey production.

Readers familiar with our story on the climate warming properties of water vapor, otherwise known as humidity, would be interested to read this thorough detail of the E.P.A.’s active avoidance of regulating water vapor as a pollutant by The Honest Broker.

Essentially, regulating humidity like carbon dioxide would also be economically destructive, and the agency seems intent on completely skirting the issue.

The Washington Post has a recent story about how the Potomac Interceptor had known issues for years with little response from D.C. Water before it broke in one of the largest sewage leaks in U.S. history.

That also tracks with our story on the ongoing bacteria issues of the Anacostia river despite billions being spent. Various reports point to leaking pipes similar to those surrounding the Potomac Interceptor being the source of bacteria on that river as well.

The gist is that, while D.C. Water is eager to spend billions on major construction projects where there is plenty of bond money is available, they seem less than eager to deal with longstanding maintenance needs.

The recent Investigative Economics story on San Francisco car break-ins quoted district attorney Jenkins saying that, of the 11,000 car break-ins during the first half of the year, few of them were prosecuted.

The story is largely about prosecution numbers being unrelated to a supposed loophole in the law, but the 11,000 total car break-ins number quoted by Jenkins deserves skepticism too.

For there to be 11,000 car break-ins in half a year, that works out to be over 60 a day (11,000 / (365/2) = 60.2).

San Francisco certainly had an issue with car theft, but 60 break-ins per day is hard to fathom.

(custom video courtesy of ozavry)

Crime in the U.S. grew concurrently as mental institutions were closing down. More patients with schizophrenia were prescribed thorazine rather than committed to an insane asylum as the drug epidemic took over modern cities. Many of them wound up in jail as prisons became de facto mental institutions that were ill equipped to deal with the flood of convicts with psychosis.

But various critics have pointed out a glaring hole in that logic: the vast majority of the institutionalized population were elderly, half of them were female. Those are two populations rarely associated with inner city crime. Mental institutions were more often used as retirement homes for those suffering dementia rather than schizophrenia.

While deinstitutionalization changed sentencing standards, courts sentencing defendants to a mental health asylum through diversion or pleading insanity has always been relatively rare. For those that are, inpatient mental hospitals still exist.

When accounting for age and gender, the inverse correlation between prison rates and mental institution rates is not as glaring. Between 1960 and 1980, the number of men aged 18 to 35—the majority demographic of those in prison—grew by about 130,000 based on Census data. That same demographic cohort for mental institutions declined only by 13,000 for the same period.

That is a sizable difference even if those suffering from severe mental illness in public are significantly more likely to commit crimes and be repeat offenders; potentially a contributing factor but difficult to describe as a main cause of prison overpopulation.

What Defines “Severe”

Additionally, the accusation that prisons were suddenly filled with the severely mentally ill is suspect as the data on mental illness in prisons is spotty at best. Much of it is self-reported on sample populations.

While self-reported data can be useful, it has major biases and definitional issues. What is considered severe is sometimes in the eye of the surveyor and the prisoner being surveyed, and the rates vary wildly from report to report.

On top of that, there are the common difficulties of statistics on psychological conditions. Clinical definitions of conditions like schizophrenia and major depression that make up severe mental illness have consistently changed over time, so even clinical surveys can be vague in historical context. While prisoners are rife with mental health issues, the number that are severe and pre-existing to detainment are hard to pin down.

A 2010 paper from researchers for the Treatment Advocacy Center and Sheriff’s Association estimated the percentage of those in jail or prison with a severe mental illness was 1.5 percent in 1930 but now stands at around 16 percent based on BOP reports. Combined with the growth in the total prison population, that would imply an explosion in serious mental illness.

Bureau of Prisons (BOP) data shows a prison population in 1933 at 189,433. With 1.5 percent mentally ill, that would be on the order of 2,850 in total.

For 2016, a 16 percent rate with around 2.2 million incarcerated would mean on the order of 352,000 mentally ill prisoners in total, far beyond what might have come from deinstitutionalization. It would put the growth rate since 1933 at a seemingly implausible 1,220 percent.

Despite coming from a large survey published by the Bureau of Justice Statistics (BJS) in 1999, the data that the 16 percent mentally ill prisoner rate comes from is based on self-reported data of whether a prisoner had a mental condition or an overnight stay in a mental hospital.

Another BJS self-reported survey of inmates in 2005 put the number of mentally ill prisoners between 45 to 64 percent of state, federal, and local jail populations. One BJS report in 2016 estimated 13 percent suffered from recent severe psychological distress.

Yet a more exhaustive, non-self-reported survey with a clinical definition of severe mental illness by the Government Accountability Office (GAO) in 2018 put the estimate at a much lower 4.2 percent of those in federal custody.

Between the 1930s and 1990s there were few exhaustive surveys, self-reported or otherwise. Most data comes from academic research sample surveys of a single prison population, like this study of inmates in Cook County, Chicago from 1983, which estimated a 6.36 rate of any severe disorder. One from 1976 from a county jail put the estimate at 22 to 23 percent, significantly higher than more recent numbers.

Without a consistent definition of “severe mental illness,” there’s no good determiner of how deinstitutionalization affected prisons.

According to current San Francisco district attorney Brooke Jenkins, the vast majority of those incidents had not been prosecuted.

“My prosecutors could go into court today even having an eyewitness to an auto burglary, but without having an individual who owns or possesses the car say they locked the doors, could be required to dismiss the case,” Jenkins said.

The city’s roughly 11,000 break-ins in the first half of this year resulted in just 86 felony auto burglary cases, about half of which actually ended in a conviction, Jenkins said. Many of those incidents happened in popular tourist destinations like the Marina District, Fisherman’s Wharf and near Civic Center.

A bill by state representative Scott Weiner would close that loophole (SB 905), something that he had been trying to pass for a few years, and it eventually passed in 2024 as section §465 and was effective starting in 2025.

While car break-ins were a huge concern for San Francisco a few years ago, almost none of the rest of that is true.

Based on numbers from DataSF, most auto burglaries have been, and continue to be, prosecuted under the section of the law that contains the loophole about requiring car doors to be locked—section §459. Over 80 percent were prosecuted in 2021 under that law, and 88 percent were prosecuted in 2025 when the new law passed.

If anything, there’s been a sharp decline in auto burglary arrests since 2019, years before the loophole was closed potentially following a sharp decline in incidence.

A San Francisco Chronicle story noted the sharp decline back in 2024 due to increased police enforcement, although that may not be totally accurate as arrests have been in continuous decline since 2020.

Similar Conviction Rate to All Burglaries

While the number of cases and arrests have plummeted, conviction rates under the old law remain about the same.

If the loophole somehow affected prosecution success—rather than whether charges were brought as Jenkins described—then conviction rates for that statute would ostensibly be low.

But conviction rates under section §459 are almost exactly in line with conviction rates for all types of burglary and robbery that aren’t automobile-related—varying between 40 and 60 percent.

Conviction rates solely refers to total convictions and does not include other potential outcomes in favor of prosecution, such as diversions or pleading guilty to a separate case. But even when considering those other conclusions, auto burglaries have had similar outcomes to other burglaries, and they are regularly prosecuted successfully.

The new law (§465) could certainly improve upon those numbers, although the data currently only shows seven cases to be prosecuted under the new law. As it was only passed in 2025, it’s still early and there is little data available on its effects.

Ongoing Prosecutorial Issues

Besides auto burglaries, the San Francisco district attorney’s office appears to have significant issues with conviction rates for all crimes as well as an increasing number of fugitives.

Robberies and burglaries used to have conviction rates almost twice as high as they currently do. The overall conviction rate used to be around 30 percent rather than current rates around 10 percent.

The number of cases left on hold because the defendant absconded and is now considered a fugitive is now around 12 percent a year when it used to be less than one percent. Then there’s a separate issue of Jenkins dropping narcotics cases.

On May 16th, Cleveland, Ohio voted to stop the $1.6 billion Slavic Village datacenter project from moving forward. The Box Elder datacenter campus in Utah has seen intense backlash, some of it over the confusion about how many acres of land it would encompass.

Criticism of the projects has largely hinged on environmental issues related to the large amounts of water used for cooling, electricity consumption, costs to ratepayers, carbon and heat generation, and their use for A.I.

The scale of the outrage targeted at datacenters appears out of place. While they use large amounts of water for cooling, so do many other industries—factories, farms, golf courses—that generate far less tax revenue for the areas they are located in.

In response, the pro-energy group American Energy Institute (AEI) published a report showing the foreign funding of nonprofits advocating against datacenter development. According to a report from Americans for Public Trust, environmental groups like 350.org and Greenpeace receive substantial sums from foreign investors like the Oak Foundation—headquartered in Switzerland and run by a British citizen Alan Parker—as well as the Swiss billionaire Hansjörg Wyss through the Wyss Foundation and Berger Action Fund, the British hedge fund manager Chris Hohn, the UK-based Quadrature Climate foundation, and the Denmark-based KR Foundation.

In February, Nebraska attorney general Mike Hilgers requested the Department of Justice investigate a collection of foreign entities funding U.S. environmental groups under the Foreign Agent Registration Act (FARA), some of which like 350.org are actively advocating against datacenters.

But besides 350.org and Greenpeace, many of the environmental groups funded by foreign entities detailed in the AEI report don’t appear to be active in anti-datacenter advocacy.

Interestingly enough another large funder of groups advocating against datacenters is the Schmidt Family Foundation. The 501(c)3 group created by Google ex-CEO Eric Schmidt heavily funds the nonprofit Food and Water Watch at the center of the activism and organizer of the letter to Congress.

The irony of course is that Alphabet, Google’s parent company, is heavily invested, if not one of the largest American investors in datacenters. Through the Schmidt Family Foundation it appears that they might be funding their own opposition.

Based on 2024 filings, the foundation holds almost $1 billion in Alphabet stock, and it gave at least $3.75 million to the signatories of the 2025 anti-datacenter letter including $1.2 million to the letter’s coordinator, Food and Water Watch along with others like Columbia Riverkeeper, FracTracker Alliance, and Physicians for Social Responsibility. In total, the Schmidt Family Foundation handed out $195 million that year to a larger number of nonprofit groups that could also be advocating against datacenters in other ways.

For example, the Google datacenter in Dalles, Oregon is at the center of a battle over water usage. The Google datacenter there currently uses in the range of 500 million gallons a year, and the city is pushing to fund improvements to the nearby Crow Creek Dam. Resident water bills could potentially double as a result.

But while Google is a large source of the town’s water consumption, the city would likely need to make the improvements no matter what, and Google already paid for a new water storage system and will likely pay millions for further upgrades including the Crow Creek Dam, as well as expenditures for ongoing water consumption.

Schmidt’s funding of FractTracker Alliance and Frack Action—two of a number of anti-natural gas fracturing nonprofits that signed the letter—is also interesting as Google is invested in numerous datacenter projects connected to natural gas plants sourcing their energy from fracking.

In 2026, Google announced the Goodnight project in conjunction with the Texas’ Crusoe energy for a datacenter in Armstrong County. Crusoe sources much of their energy from methane flaring and hydrologic fracturing, originally as a way to fund bitcoin mining. Google is also connected to other datacenter projects with attached natural gas plants in Illinois and Nebraska.

A previous Investigative Economics story detailed how Google funded the opposition to a nuclear powered datacenter in Georgia despite Google itself involved in nuclear-powered datacenters in other parts of the country.

But the pound was overvalued, and, despite its best efforts, the Bank of England couldn’t prop up its value any longer. Currency speculators swooped in to bet against the pound, with George Soros’ Quantum Fund alone betting an incredible £10 billion on its collapse by borrowing and selling 61 percent of the pounds in circulation at the time (£16.4 billion). That bet would effectively earn him £1 billion in profit—possibly one of the largest single day windfalls in investment history. Britain’s attempt to join the EMS was abandoned and within a month sterling would lose 20 percent of its value.

Campaign to Drive Sterling’s Value

In their efforts to prop up the currency, the Bank of England borrowed at least £5 billion European Currency Units (ECU)—a basket currency that was based on a collection of various European currencies before the Euro—to buy up pounds and keep its value high as well. Soros was quoted saying that the British finance minister was willing to borrow $15 billion to defend sterling. Interest rates would spike to an incredible 27 percent for the day.

Likely due to that debt, total liabilities for the bank would double within a week, topping £10 billion at a time. While some interviews have described the attempts to prop up sterling as completely ineffective, the value against the dollar would rise by 5 percent in the week leading up to Black Wednesday.

But the fervent effort by the bank to prop up the value of sterling was only temporary, and the U.K. would announce its exit from the ERM as the pound went into freefall.

Yet somehow borrowing billions and buying billions of pounds that would lose 20 percent of their value in a matter of days didn’t make much of an impact on the bank’s bottom line. A loss of £1 billion to currency speculators would seemingly stand out in total liabilities. Other anecdotes have put the price at £3.3 billion. But within a month total liabilities were back where they were pre-crisis. It was as if the bank borrowed billions only to immediately pay it right back with no major loss.

Large holdings of currency—whether it be in pounds, ECU, or another currency that the U.K. might have traded for—would ostensibly show up in the total reserves of the bank, but there is no visible sign of billions in debt or reserves left on their books when it was all done.

Devaluation Was Expected

While sometimes Soros’ bet is described as investing genius, the collapse of the pound was not that big of a surprise. Italy and Spain also devalued their currency around the same time when they realized they couldn’t keep up their currency’s value and the ERM was not in their best interests.

For years prior, Britain struggled with the burden of high exchange rates, which made their exports more expensive to the rest of the world. Additionally, high interest rates burdened ordinary borrowers, from local businesses to those trying to buy a house. The idea that Britain would continue to keep their currency value high by raising interest rates even higher just to be a part of Europe comes across as economic suicide.

Some of their strategy was based on the hopes that Germany’s Bundesbank would lower their interest rates, which would in turn cause the pound to be more valuable. But in the days prior to Black Wednesday Germany announced the increase to the country’s interest rates, and it was a measly .25 percentage points for the Lombard rate and .50 percentage points for the discount rate. With that announcement, the collapse of the ERM agreement was all but inevitable.

The recent change was ostensibly an indication of Jenkins’ tough on crime approach to the office: she was still filing charges at a similar rate yet winning more cases.

While all of that was true at the time the article was published, her office has seen a steep drop in the number of narcotics cases charged during her tenure since then. What was a 73 percent prosecution rate for narcotics crimes is now down to 50 percent. In 2023, 408 cases—25 percent of narcotics cases that year—were discharged, meaning charges were dropped before being filed. While that has since gone down, it’s an astounding amount.

Rather than prosecuting new cases, the district attorneys’ office is now more likely to move-to-revoke (MTR) cases—which means revoking a prior probation and moving forward on the prior charges that the probation was set under—or referring charges to other agencies.

The change is likely related to how the police—mainly San Francisco Police Department (SFPD) but also other entities like the county sheriff’s office and Bay Area Rapid Transit (BART)—were making 983 more narcotics arrests a year in 2025 compared to 2022.

Improving Conviction Rate

Prosecution rates in San Francisco hit a historic low in 2013. While that was under George Gasçon who took the reins in 2011, the trend started under the previous district attorney Kamala Harris who dropped plenty of cases, particularly those related to fraud, theft, and counterfeiting. Under Gasçon is when the trend would reverse itself.

While Gascon took on more cases, he failed to convict on many of them, instead preferring diversions. While Jenkins declines to prosecute many narcotics cases, conviction rates have significantly improved under her tenure.

Essentially, the complete collapse of the DA’s conviction rate largely under Gasçon in favor of diversions has flipped. At a time felony conviction rates barely made it above 55 percent but have since risen above 70 percent.

But maybe the biggest unexplained shift is the large jump in the disabled population. Starting in 2020, an additional 1 million people over 16 would be identified as disabled per year. There are now 6 million more people disabled than there were pre-pandemic. What was once 9 percent of the population would now be almost 11 percent.

Potentially the large swing could be some remnant of those affected by COVID-19 or appealing to COVID-19 related benefits, but detailed survey data from the Census shows distinct growth across practically all types of disabilities including hearing, cognitive, and vision difficulties since 2019 to 2020.

Much of that is likely a remnant of the elderly, whose population is rising with generational trends. The majority of population growth in disability happened in those over 64. The pandemic might have simply encouraged those to seek diagnosis during a time when not much else was happening. As a percentage of the population—the prevalence—it actually went down for those over 64.

Instead, one of the largest growing prevalent categories would be those with a cognitive difficulty between the ages of 18 and 34 based on U.S. Census data.

There are now 3 million more people reporting to have a cognitive disability since 2016, with half of those being between 18 and 34—going from 4 percent of the population to 6 percent.

Anxiety, Autism, and ADHD

While the change surrounding the COVID-19 pandemic is unique, the growing prevalence in mental health disorders is nothing new.

Prior stories from Investigative Economics noted the sharp upswing in reported anxiety disorders, particularly in white women in various states like Minnesota.

Based on Social Security data, fewer workers are filing for disability claims for themselves, but significantly more are filing on behalf of their children, particularly for mental health conditions.

Autism Growing On Top of Definitional Change

There has already been an acknowledged growth in certain mental disorders due to revised definitions from the updated DSM V manual, like that of autism and ADHD. But those came out in 2013.

And while rates of autism and ADHD had been rising following those definitional changes, the post-pandemic changes made their prevalence grow even faster, and not by all age groups at once.

Based on sample data from the National Survey of Children’s Health (NSCH), there would be a sharp upswing first in those aged 3 to 5 years in 2020. Then by those 6 to 11 years. And then a few years later kids 12 to 17 years old would show higher rates of autism beginning in 2022 to 2023. Rates for ADHD diagnosis would be similar.

• Senate Report on BCCI

• House Banking Report on BCCI

• The U.K.’s Sandstorm Report on financial audits of BCCI (via Internet Archive), Wikileaks Version

• Declassified C.I.A. Files on BCCI

• False Profits: The Inside Story of BCCI, The World’s Most Corrupt Financial Empire (book)

Miscellaneous Articles on BCCI (plenty more not included here):

• New York Times

• Foreign Policy

• Washington Post