American aviation's last line of defense is quickly evaporating. Disgruntled federal air marshals, according to National Review, are leaving the service at an alarming clip — in 2014, an average of 10 agents were reported to have left the service's Washington, D.C., office every month. It seems only a matter of time until the U.S. government will be forced to consider a major overhaul of one of its biggest post-Sept. 11 security upgrades.

That could be a good thing, but only if Washington is willing to entirely rethink the program. There's little reason to believe that a Federal Air Marshal Service — which is currently thought to have around 3,500 officers — was ever the best way to protect the more than 26,000 commercial flights that take off in the United States, on average, every day. America's planes can be made far safer, far more efficiently.

The idea that armed guards should protect passenger flights dates back to the administration of President John F. Kennedy, who placed border agents onto jets in response to a series of hijackings of planes by Cubans. The program was formalized as a Sky Marshal program in 1968, and by 1970 there were 1,784 active officers. Nonetheless, between 1968 and 1972, approximately 75 U.S. flights were hijacked, according to ProPublica — a significant increase over the hijack-happy first half of the 1960s.

Hijackings only declined after airports introduced technologies like metal detectors and X-ray machines in the mid-1970s, and the U.S. and Cuba implemented anti-hijacking laws. The U.S. government soon drew the reasonable conclusion that air marshals (or sky marshals, as they were then known) weren't terribly useful, and the best place to ensure airplane security was on the ground before takeoff. In the following decades, their ranks shrunk dramatically — by 9/11, the service had only 33 agents.

Of course, those numbers expanded very quickly thereafter. Congress, with the approval of the Bush administration, increased the air marshal program's budget from $4.4 million in 2001 to $545 million in 2003. An agency that previously had one office suddenly expanded to 26 field offices. Previously, managers made individual decisions about placing agents on planes; now they had access to an automated system designed to assign thousands of agents to high-risk flights, including international routes.

But amid the frantic post-9/11 push to strengthen security, the government neglected to scrutinize whether a major ramping up of the Air Marshal Service made much sense. The program now costs $825 million per year. It's only fair to wonder whether the money spent hiring air marshals — who, in aggregate, averaged only 4.2 arrests per year during the 2000s — could be better spent on other, lower-cost safety measures.

Strangely, Congress and the rest of the federal government have never taken much interest in assessing the program's value. In January, Sen. Tom Coburn, the former chair of the Senate's Homeland Security and Governmental Affairs Subcommittee, published a report earlier this year emphasizing that the government hadn't even bothered to compile the relevant data:

In part due to limited publicly available oversight evidence, it is unclear to what extent the program is reducing risk to aviation security, despite the more than $820 million annually that is spent on it.

Thankfully, others have taken the limited amount of unclassified information available and have issued assessments of their own.

The most widely cited cost-benefit analysis of the program was conducted by a U.S. political scientist and Australian engineering professor in 2008. They concluded that the costs imposed by the Federal Air Marshal Service — they estimate an annual cost of "$180 million per life saved" — are unreasonably high. The government's money, they argue, has been far better spent on another ongoing safety measure: hardened cockpit doors. According to their analysis, those doors have an annual cost of $800,000 per life saved.

They suggest other measures might be similarly efficient. One possibility would be for airports to focus on improving pre- boarding security. Another would be to place more trust in the ability of passengers to serve as a last line of defense. Since 9/11, passengers have managed to subdue a fair number of in-cabin disturbances initiated by unruly passengers and crew.

Of course, an air marshal can do things that a cockpit door and unarmed passengers can't — like shoot people. But if the goal is to have an armed presence on flights, the federal government already has a far cheaper option: the Federal Flight Deck Officers program, under which eligible flight crew members are trained by the TSA to use firearms aboard flights. It's been around since 2003 and, according to the Air Line Pilots Association, which supports the program, FFDO costs approximately $13 per flight — compared with the air marshal program, which costs around $3,000.

In fact, if anything calls into question the cost-effectiveness of the Air Marshal Service, it's the 14.6 percent cut its budget has taken since 2012. Those cuts, according to the Department of Homeland Security, had led to elimination of air marshals "by attrition" — yet so far, at least, there's been no increase in terror on American airplanes.

That's not an argument for abolishing the Federal Air Marshal Service entirely. It still has a role, especially on particularly risky routes where there may not be FFDO crew, or in accompanying dignitaries. But it's time to get past the idea that America's safety is being compromised by marshals' shrinking numbers. In all likelihood, it makes no difference — except to the wallets of U.S. taxpayers.

Adam Minter is based in Asia, where he covers politics, culture, business and junk.