NEW YORK CITY – It was one of the deadliest fires in New York City in recent years.
Guatemala-born Miguel Chan was so desperate to save his family that he threw his infant daughter out the window of his burning Brooklyn apartment. She survived, others did not.
"I lost my wife, my four friends," Chan says of the Jan. 30, 2010 fire. Police immediately suspected arson and three days later arrested Daniel Ignacio, a neighbor who confessed he was drunk and influenced by "demons or devils" when he set the apartment fire killing five Guatemalans.
It was an open-and-shut case of arson to everyone – except the federal government.
Like 99 percent of New York's arsons, the intentional fire that killed Miguel Chan's wife and his friends was never reported into the federal database that tracks arson in America.
"This is unbelievable. Everyone knows that this was an arson," said the Rev. Erick Salgado, pastor of Chan's Guatemalan church who helped police make an arrest.
Just 5 percent of all residential building fires are intentionally set, according to the National Fire Incident Reporting System – or NFIRS – the world's largest, national database of annual fire incident information and part of the Department of Homeland Security.
But arson actually is much more common than is reported by the U.S. government, Scripps Howard News Service found in a yearlong national investigation. Most acts of arson in America go unreported to the federal government, the investigation found.
For example, in 2011:
-- Chicago reported just 61 building arsons when it had at least 192.
-- Houston said it had 25 intentional fires when it really had 224.
-- Indianapolis reported no arsons when it should have reported at least 216.
-- New York told of just 11 arsons instead of 1,347 it really discovered.
In all, Scripps contacted 10 fire departments in America's largest cities to ask for case-by-case records of their arsons to compare what was reported to NFIRS against what should have been reported.
One city, Detroit, said it could not produce a complete arson count or make its records available because the financially-strapped city lacks the personnel to do so.
"This is a modest guess, but I think 75 percent of our fires are arson, maybe more," said El Don Parham, Detroit's chief of fire investigations. "We are not able to cover (investigate) even half of our fires. We have to prioritize."
The nine remaining cities did report data, although New York has yet to give the exact locations and dates of its arsons. These cities originally reported 652 arsons to the federal government in 2011, but actually detected at least 2,754 deliberately set fires that year.
That means three-fourths of the arsons uncovered by investigators in those cities went unreported to the U.S. Fire Administration, masking a major threat to public safety.
"Arson is grossly underreported," concluded Bill Degnan, New Hampshire's fire marshal and the president of the National Association of State Fire Marshals. "I believe the rate of arson in America is somewhere between 40 percent and 50 percent, in that range."
The quality of the federal data is haphazard and the underreporting of arson could have serious consequences:
-- Cities and towns are not focusing enough of America's 1.1 million firefighters to combat arson.
-- Up to half of the 3,000 fire deaths each year should be treated as homicides.
-- Arsonists may remain at large, free to burn.
-- Much of the $15.5 billion paid last year by insurance companies (and their clients) should be contested since arson often involves fraud.
The U.S. Fire Administration has given out $4 billion in grants to participate in NFIRS without penalties for inaccurate information. The agency does not audit the reporting.
The Scripps investigation found that every major city it contacted has failed to report significant numbers of arsons. And there are many reasons why.
The worst was New York City, which in 2010 reported only 19 acts of arson in building fires even though its Bureau of Fire Investigation found there were at least 1,486 intentional building fires that year.
City fire administrators say it's a technology issue and deny there was any motive in not reporting the 2010 fire at Miguel Chan's apartment as arson.
"This was an administrative and clerical thing. We aren't trying to keep anything quiet or secretive," said Jim Long, spokesman for the Fire Department of New York. "We were holding press conferences in which we identified that as an incendiary fire."
The city's Bureau of Fire Investigation actually detects arson in about 40 percent of the cases it investigations in recent years. Detectives at the Bureau of Fire Investigation have their own reporting system which does not have a physical link to NFIRS.
"We are working toward linking those two databases," Long said.
The Houston Fire Department reports only about one in every eight arsons it detects to federal authorities. In 2011, according to the Scripps review, it detected at least 224 arsons, but reported only 25 of them to the U.S. Fire Administration.
As was common with many of the departments contacted in the investigation, Houston officials said there is disconnect between the firefighters who battle blazes and the arson investigators who search for their causes.
"It's the firefighters on the trucks who are filling out the NFIRS reports. We don't," said Deputy Chief Ed Arthur, head of Houston's Arson Bureau. "But they are not going to know what cause the fires. We make that determination."
Degnan said the National Association of State Fire Marshals is aware of the problem and established a group a year ago to investigate why most arsons go unreported. The group will make recommendations in January calling for the nation's fire departments to "close the loop" and report the outcomes of their arson investigations, he said.
He also said local investigators often are reluctant, even fearful, to report arson to NFIRS.
"Unfortunately, many people shy away from making the correct call even though they might believe that a fire is incendiary," Degnan said. "They are concerned that they are going to be ‘called in' if they call it wrong."
Firefighters have told his group that "making the call" – declaring the cause of a fire – means putting their reputation on the line with their chiefs. Many don't want to risk making a wrong call. That means a lot of fires go down as "cause undetermined" or permanently "under investigation."
Officials with the U.S. Fire Administration have declined repeated requests for interviews, including requests via certified mail.
John R. Hall Jr., research director at the National Fire Protection Association, a non-profit group that sets investigation standards, defends the low arson estimates.
"Over the years I've had fire chiefs and fire officers tell me that they think the arson numbers are underreported. They've been of the view that it's half or more than half," Hall said. "They always assume they know better than the data."
Most arsons in Houston, as in New York, are reported to NFIRS as "under investigation."
Among the fires still waiting for resolution more than two years later is the much-publicized Jan. 3, 2011 blaze that destroyed a $2 million Victorian-style home in Houston's trendy Heights neighborhood that killed Evelyn "Patti" Worthington, 68, and her caretaker. The city arson squad still lists the cause as "pending."
"Obviously, it would be good for everyone if there was a final resolution to this," said Worthington's ex-husband, David Worthington of Naples, Fla. "Was it an accident? Possibly. But it's hard to surmise what happened."
One of Worthington's neighbors, Bart Truxillo, isn't surprised fire investigators are slow in making a call in the Worthington fire since he, personally, saw them call another one wrong. Fire investigators at first blamed faulty wiring for two deaths in a home Truxillo owned until an insurance investigator proved that Annie Caballero, 22, had died before the fire started 13 years ago. The case is now considered an unsolved homicide.
"All the fire department did was to bring their dog to sniff around, but they didn't find anything," said Truxillo. "They were going to take the easy way out. Thank God for that insurance investigator."
Scripps also found serious technical problems that distorted arson information in many fire departments. The Colorado Springs Fire Department, for example, reported all 677 fires as "cause undetermined" from 2009 to 2011, even though its arson unit has identified at least 273 intentional fires. Authorities there have promised to fix the problem.
The worst such glitch was at the Indianapolis Fire Department, which reported to NFIRS only a single arson over a six-year period, even though the arson unit in the nation's 12th-largest city finds 400 or more deliberate fires each year.
Indianapolis opened its books to the Scripps review, which found that 32 percent of all fires causing at least $5,000 damage in Indianapolis were intentionally set. The city had reported an arson rate of zero for many years to federal authorities, who never challenged the data.
Gregory Gates, battalion chief of special operations in Indianapolis, said he believes the city's arson rate is even higher because many intentional fires are missed. The city, like most communities, lacks sufficiently trained investigators to probe every fire, he said.
"Arson for us is somewhere between 50 percent and 70 percent," Gates said. "We have as many fires that are listed as ‘undetermined' that are actually arsons as we have that are listed as arsons."
The consequences of improperly detected and reported arsons are severe, fire officials warn.
Nearly 3,000 Americans die each year in fires. Only about 300 of those deaths are attributed to arson, according to the U.S. Fire Administration, and, therefore, are legally regarded as homicides. But if arsons actually account for up to half of the nation's fires, hundreds of homicides could go undetected and uninvestigated each year.
"Unfortunately, if the fire is not fully investigated and the scientific method of fire investigation is (not) used to determine the origin and cause, a person could go free because of that not being reported correctly," New Hampshire's Degnan said.
Homeowner insurance fire losses in 2011 were about $15.5 billion, accounting for 27 percent of all insurance payouts.
"Arson has always been around and always been a concern. But we haven't heard that there is significant underreporting," said David Corum, vice president of the American Institute for Chartered Property Casualty Underwriters based in Malvern, Pa., and head of its research arm, the Insurance Research Council, which is funded by the nation's largest insurance companies.
Corum said his group will begin a study into unreported arsons.
"If it really is 40 percent or 50 percent, well, that is shocking. That would be a real wake-up call," Corum said. "If these numbers are correct, then there is a significant problem with awareness and attention on this issue. We will take a look at this."
Degnan and others warn underreported arson means America's policymakers are not allocating money and manpower correctly to suppress fire since intentional fire setting is not a priority in most communities.
One recent federal study from the National Institute of Standards and Technology and the U.S. Forestry Service suggests that America's arson problem – if undercounted and ignored – could worsen any community's decline into social disorder, an effect sometimes called the "Broken Windows" theory.
"The consequence of not having the right measurement of arson is that we underappreciate the size of the problem," said NIST economist David Butry, one of the study's authors. "That certainly would affect how it is prioritized by society."
Their study concluded that "crime prevention and urban revitalization programs may be as valuable as fire suppression and law enforcement" in fighting arson.
Until NFIRS was created in 1977, there had been no formal gathering of U.S. fire statistics. As an incentive for fire departments to participate in the program, the U.S. Fire Administration has paid more than $4 billion in Assistance to Firefighter Grants since the program was started more than a decade ago. The grant program does not stipulate that information must be accurate for a department to receive the grants.
Brad Pabody, chief of the National Fire Data Center, wrote in an e-mail in April to say the federal administration "does not audit NFIRS reports" because of resource limitations.
(Reach Scripps national reporter Thomas Hargrove at firstname.lastname@example.org.)