This guest article is from David Delaney, President & CEO, at Insight Analytics.
The earthquake that delayed a World Series lasted about fifteen seconds. The losses for insurers who'd written the peril out of their policies didn't stop when the shaking did.
At 5:04 p.m. on October 17, 1989, fans at Candlestick Park were waiting for Game 3 of the all-Bay Area World Series when the ground started moving (MLB.com)¹. A magnitude 6.9 earthquake, centered in the Santa Cruz Mountains near Loma Prieta peak, about 60 miles southeast of San Francisco, had ruptured a segment of the San Andreas fault zone (USGS)². ABC's live pregame broadcast carried it to tens of millions of viewers, and Al Michaels got out "I'll tell you what, we're having an earth" before the picture cut away (ESPN)³. The Series didn't resume for ten days.
ESPN's oral history of the earthquake that interrupted the 1989 World Series, featuring Al Michaels, Joe Torre, Tony La Russa, and others who were there. Source: ESPN (YouTube).
The game was a footnote. Unfortunately, what the quake exposed about earthquake insurance wasn't.
This year's World Series opens October 23 in Milwaukee with no California team in it (Wikipedia)⁴, but when it comes to the earthquake risk under San Francisco and Los Angeles, THAT doesn't wait for a baseball schedule.
Aerial view of the collapsed Cypress Street Viaduct, Interstate 880, Oakland, after the October 17, 1989 Loma Prieta earthquake. Photo: H.G. Wilshire, U.S. Geological Survey.
Total economic damage ran roughly $6 billion by USGS's accounting. Insured losses were far smaller: $960 million in 1989 dollars, about $2.48 billion in 2025 dollars, across 198,920 claims, making Loma Prieta the third-costliest U.S. earthquake by insured loss (Insurance Information Institute)⁵. That was only about 14% of the roughly $7 billion in property damage counted at the time (Los Angeles Times)⁶. Most of the cost fell on uninsured owners, lenders, and the public sector.
The earthquake killed 63 people. Forty-two died when the double-decked Cypress Street Viaduct in Oakland collapsed (NIST)⁷. In San Francisco's Marina District, shaking liquefied bay landfill and snapped gas and water lines, and a ruptured gas main started the worst fires (SF Fire Department)⁸.
Earthquake insurance take-up was already low: roughly 20% to 28% of California homeowners in 1989 (Moody's RMS)⁹. In 2025, 12.48% of California residential policies carried it (California Department of Insurance)¹⁰. The gap hasn't closed.
Most property insurers exclude earth movement outright, which makes a major earthquake look like someone else's exposure. Loma Prieta shows the exposure runs through lines most insurers already carry.
Fire following. California law has required insurers to cover fire that follows an earthquake since 1906, when fire burned far more of San Francisco than the shaking did (Insurance Code §2081)¹¹. The Marina fires were exactly this loss: shaking excluded, fire covered.
Water damage. Sprinkler failure drove more than half of FM Global's recorded earthquake losses (FM Global)¹². Homeowners forms carve back fire following an excluded peril, but generally not water. Commercial buyers can add it back only if they asked before the quake.
Claims expense. Every claim needs adjusters and engineers to sort shaking from fire from water, even when it ends in a denial.
Commercial ground shaking. Commercial carriers write shaking back in more often than homeowners carriers. Take-up runs an estimated 15% to 20% for Bay Area commercial properties and 30% to 40% for the largest urban risks (Moody's RMS)⁹.
Workers' compensation. Benefits are set by statute, so there is no earthquake exclusion and can't be one. One modeled repeat of 1906, striking during working hours, put comp losses near $3 billion (TransRe)¹³.
Liability. California paid $71 million on 335 claims over the Cypress Street and Bay Bridge collapses under a special no-fault statute (Los Angeles Times)¹⁴. A structure doesn't need earthquake coverage to generate an earthquake claim against whoever owned, designed, or built it.
USGS shaking-intensity maps comparing the 1868 Hayward Fault earthquake (left) and the 1989 Loma Prieta earthquake (right), with the 1989 epicenter marked. Map: U.S. Geological Survey.
USGS puts the 30-year odds of at least one magnitude 6.7 or larger earthquake in the Bay Area at 72% through 2043 (USGS)¹⁵. The Hayward Fault last ruptured in 1868, 158 years ago, against an average gap of about 140 years between its major earthquakes (USGS)¹⁶. Southern California's odds are 93%, and the southern San Andreas segment that broke in 1857 has gone 169 years without a repeat, against an average interval of roughly 100 years at one well-studied site (USGS)¹⁷.
The science needs precision. USGS describes "elapsed time exceeding the average recurrence interval," not an imminent earthquake, and some seismologists object to how loosely "overdue" gets used (Temblor)¹⁸. The defensible claim: the strain is real, and on at least two faults it has outlasted the historical gap between major releases.
USGS's HayWired scenario, a magnitude 7.0 on the Hayward Fault, puts combined property damage from shaking, fire, liquefaction, and landslide at roughly $74 billion (USGS)¹⁹. Fire following adds roughly 450 large fires and close to $30 billion more (USGS)²⁰. CoreLogic put total damage near $170 billion and estimated only about $30 billion of it, under a fifth, as insured (USGS)²¹.
The ShakeOut scenario, a magnitude 7.8 on the southern San Andreas, models roughly 1,800 deaths and $213 billion in economic loss (USGS)²². No modeled insured-loss figure has been published for it, so none is cited here. Both scenarios are 8 to 18 years old and almost certainly understate a repeat today.
A clean earthquake exclusion on the property form isn't a clean exit. Fire following, water damage, claims expense, workers' comp, and liability all travel on policies already in force, in lines never priced with a major Bay Area or Southern California quake in mind.
The protection gap HayWired modeled, roughly 80% of total damage uninsured, is the same dynamic flood exposed after Superstorm Sandy. The peril where the exclusion looks cleanest on paper is often where the uninsured loss, and the political pressure that follows it, ends up largest.
Loma Prieta delayed a World Series for ten days. The faults involved haven't rested in the 37 years since.
Figures were cross-checked against primary sources on 2026-10-18, including USGS, NIST, the Insurance Information Institute, and the California Department of Insurance. Caveats:
.jpeg?width=120&height=120&name=Delaney%20David%20headshot%20(from%20LinedIn).jpeg)
About the author: David Delaney
David Delaney is President & CEO of Insight Analytics, an insurance analytics and underwriting technology company focused on real-time underwriting insights and catastrophe risk analytics. He has more than 25 years of experience across insurance, reinsurance and InsurTech, and previously held Vice President roles in catastrophe risk management and catastrophe underwriting at AIG. Connect on LinkedIn.
5 min read
Following a Hurricane’s Landfall, What Can Be Known About Insured Losses? A Primer for Insurance Executives by Mark Szretter (Para Español – ...
10 min read
.jpg?width=30&name=Nick%20Lamparelli(Headshot%20GS).jpg){kind=small}
Nicholas Lamparelli : Dec 4, 2017, 12:00:00 AM
Profiles in Risk – E44: Cuthbert Heath & The Making of the Modern Lloyd’s of London…”Why Not?” by Nicholas Lamparelli Audio Player ...
4 min read
DMCOs – What are they and why they will ultimately replace all but the truly exclusive leads Vendors? by Carl Moulton It is said that necessity...
Insurance Nerds