NEW YORK — To an outsider, military pensions sound ridiculous: You can put in 20 years starting in your late teens or early 20s and “retire” at the time when most people are hitting their peak earning years. Obviously, that’s a very expensive benefit for the government to provide. Should we ratchet up the retirement age? Some economists think we should.

People supporting the status quo will probably argue that the military is more physically demanding than most jobs, and therefore you have to expect people to retire earlier. But the pension is available to everyone in the military, not just infantrymen. Moreover, it is disproportionately used by officers, not enlisted men, and by the time they have 20 years in, officers are spending a lot less time hauling heavy things and running around in the mud.

But there’s another problem with rejiggering the Army’s pension schedule, and that’s the way it interacts with the “up or out” system that the military uses for officers’ careers. Basically, officers who don’t get selected for promotion get fired.

The military is not the only institution that uses this method. It’s also popular with consultancies, law firms and investment banks. That system is archaic and barbaric, and whatever it gains you in reduced payroll costs, it loses you in accumulated human capital, and it also earns you a back-stabbing corporate culture.

Of course, no one asked me, and I expect that those sorts of firms will continue to use up-or-out pyramids for the foreseeable future. But what do all these firms have in common with each other, and not with the military?

They pay really well. The senior people who survived the tournament get paid even better, of course. But even the entry-level jobs pay better than most of the alternatives.

The opposite is true in the military. It pays badly in the beginning and it pays badly at the end, relative to what those folks could have been making if they’d been steadily moving up through the ranks in private industry.

There are all sorts of ancillary benefits, of course, but also all sorts of ancillary costs, such as the fact that your employer expects you to pick up and move your whole family somewhere else every few years – which is not just inconvenient but, in this modern day, plays hell with the career prospects of your spouse.

Military pensions are extremely generous compared to those in the private sector, but without them, we might have to pay more, or watch the quality of the mid-career officer corps decline. Note that the biggest industry that uses the up-or-out system without paying big salaries – academia – makes up for it by giving the winners a paycheck for life, which is even more extreme than a military pension.

I’m against tenure, and I’m generally also against backloading compensation in the form of defined-benefit pensions. These systems are terrible for workers who don’t stay long enough to qualify for a pension, and they’re not good for the workforce as a whole, either: They encourage risk-averse time serving at the expense of your organization’s mission, not to mention clogging your organizational chart with burned-out characters who ought to go do something else but won’t leave until they get that golden ticket.

But I also recognize that de-emphasizing backloaded compensation means that you have to pay more up front, right now. No, don’t wave around actuarial charts showing that your new plan has just as high an expected value as the old one; the workers won’t buy it, and I’m not buying it, either. You’re asking them to take more risk, and just as in financial markets, you’re going to have to pay them to do so. I think that’s a good idea – that ultimately, doing so creates more value for everyone.

What you can’t do is tinker with one part of the system and expect all the other parts to stay the same. If we move the nation’s military members off the current defined-benefit system, we are probably going to have to make up for it in some other way. Especially once the nation’s politically powerful veterans groups get involved.