Deposit insurance coverage expanded over time from a temporary small-deposit measure to effectively blanket protection—statutory limits rose to $100,000 (1980) then $250,000, but in practice ~99.7% of all deposits ended up insured, and innovations like CDARS let wealthy depositors swap a $50 million deposit across ~200 banks to insure the entire amount, completing the destruction of risk-based depositor discipline.

Roughly half of all subprime losses are attributable to no-doc lending's adverse-selection problem (not verifying borrowers attracts bad credit) and the other half to the modeling assumption that housing prices could never fall, which was built into stress tests; both assumptions were knowably unreasonable in advance.

The real-estate bubble was not a US-only phenomenon, which undercuts the objection that Fannie/Freddie cannot explain global bubbles; many governments over-promoted home ownership in the prior 15 years, and Norway had a Fannie Mae-style institution before its pre-WWI crisis.

Deposit insurance protects small banks by removing competition along the dimension of risk: without it, large stable branch banks (like Canadian banks) could win depositors by advertising their stability, but government backing of deposits eliminates that competitive advantage, which is precisely why small banks lobbied for it.

When deposit insurance's perceived reliability falls, market discipline returns: in Mexico's 1990s crisis, nominal 100% coverage remained but depositors, doubting the government could cover the losses, withdrew from weak banks—demonstrating that depositors discipline banks whenever they actually bear risk.

Fannie Mae and Freddie Mac deliberately entered subprime no-doc lending in a big way in 2004 over the explicit objections of their own credit-risk managers (who knew from late-1980s experiments that not verifying borrowers attracts bad credit risks like a magnet); as the dominant 800-pound gorillas, their decision to 'make markets' caused subprime no-doc volumes to triple in 2004.

Securitization itself is not the flaw: credit-card securitization used the same originate-and-distribute model for 30 years without failing alongside subprime, because the prudent incentives long applied in credit-card securitization were deliberately set aside in mortgage securitization—made possible by government programs.

Government protection policies—deposit insurance and bailouts—plus government policies promoting real estate development (analogous to Fannie/Freddie) are the key factors that changed between the two eras and that invite excessive risk by removing market discipline, since banks then take risk at someone else's expense.

Canada, with a branch-banking system from the 1860s and no central bank until 1935, experienced none of the panics the US suffered (1873, 1884, 1890, 1893, 1896, 1907), demonstrating that US panics stemmed from the fragility of its unit-banking structure rather than from an inherent need for central banking.

Deposit insurance was historically known to be a bad idea—FDR opposed it because all eight state deposit-insurance systems created in the 1910s-1920s failed disastrously with worse bank losses than uninsured systems—yet it passed in ~1933 as a temporary emergency measure with very limited coverage, pushed by Henry Steagall of Alabama as special-interest legislation to protect politically influential small rural banks from competition along the dimension of risk.

The recent crisis wave required two coinciding trends: expanding deposit-insurance/bailout coverage AND increased macroeconomic risk—the low-volatility 1950s-60s offered little risk to bet on, but rising inflation and business-cycle volatility from the 1970s provided risk to exploit, so the S&L crisis emerged in the late 1970s as expanding coverage met expanding risk.

Even Friedman and Schwartz endorsed deposit insurance in 1963, but they were observing an unusually placid, limited-coverage period and were wrong; deposit insurance is unnecessary because small savers could be protected via postal savings (invested in Treasuries with no risk), and it has been the single most important contributor to financial-system risk by removing risk-based bank competition.

Britain's recurring 19th-century banking panics (1819-1857) stemmed from Parliament pressuring the Bank of England to provide an implicit put option on the London bills market—a quid pro quo for its privileges, directly analogous to Fannie Mae/Freddie Mac—and the crises ended only after the Bank publicly renounced this put in 1858 and credibly enforced it by letting Overend Gurney fail in 1866 despite having bailed it out in 1857.

The current US system has three compounding failures: deposit insurance and too-big-to-fail (first seen in the 1983-84 Continental Illinois bailout) removed depositor discipline, leaving only prudential regulators to identify risk; those regulators fail because they outsource risk measurement to the banks themselves and to ratings agencies; and the government simultaneously subsidizes huge housing risk—so when loose 2002-2005 monetary policy (negative real Fed Funds, >1-2% departures from the Taylor rule) added cheap credit, a severe banking crisis was the predictable result.

Banking crises share a cyclical timing driven by loose monetary policy or credit cycles, but timing alone is insufficient—a crisis of large magnitude requires something deeply wrong with the microeconomic incentives in the banking system, not merely a normal business-cycle reaction.

Government housing policies—FHA lending pressures, the affordable-housing push on Fannie Mae and Freddie Mac since ~1994, and a 2006 law plus 2007 SEC rules making it harder for ratings agencies to be tough on subprime MBS—distorted incentives by inducing lending to poor-credit borrowers with no money down and no documentation, which is a large part of why the US housing sector failed so severely.

The Minsky/'madness of crowds' explanation—that banking crises arise from unchanging human frailty, greed, or panic—is refuted by the basic fact that crisis frequency varies enormously across time and place; if human psychology were the driver, we would not see such variation, so something other than human nature must explain the difference.

During the first globalization era (1874-1913)—free banking entry, fixed exchange rates, large capital flows—there were only about four major banking crises (Argentina 1890, Australia 1893, Norway 1900, Italy 1893), all sharing government-policy distortions, whereas from 1978 to the present there have been roughly 140 major banking crises, with about 20 of them larger than the two biggest pre-1913 crises.

Banking panics and waves of large bank failures are distinct phenomena that only sometimes overlap; the Fed's 1913 founding genuinely reduced US panics by stabilizing seasonal reserve fluctuations (tied to the cotton market) made acute by America's peculiar unit-banking structure, but the US did not historically suffer the large-loss bank-failure crises seen elsewhere, so the Fed's panic-prevention role is separate from the over-speculation/large-loss problem.

The most shocking fact of the crisis is that even after mid-2006—when ratings agencies' own assumptions were being violated and delinquencies in the 2004-05 cohorts were forming, prompting Deutsche Bank and Goldman Sachs to protect themselves—Fannie, Freddie, Citibank, UBS, and Merrill Lynch continued to originate, sponsor, and hold subprime deals at peak levels; had just those five institutions stopped in mid-2006, the crisis would not have occurred.