I know a building with four apartments on the top floor.

Years ago, one of the owners started having problems with the roof. Leaks.

A team of roofers came. They repaired it. It leaked again. Another team came. Same story.

At some point even the professional building manager seemed to more or less give up. The roof wasn't collapsing. The problem wasn't catastrophic. Somebody would patch something when necessary, and everybody carried on with their lives.

Years passed. Eventually, the building changed managers. The new manager looked at the situation and said something quite different.

He didn't claim he knew the magical repair the previous roofers had somehow missed. He basically said: "This roof is going to become more and more of a problem. Eventually, we're going to have to replace it."

Fair enough. But then came the interesting part. He wasn't proposing that the building replaces it. He was proposing that apartment owners start paying for it. Now.

His idea was simple: create a dedicated fund for the roof. Everybody contributes a relatively small amount over the years. Maybe the roof lasts another ten years. Maybe less. But whenever replacement finally becomes unavoidable, the building has money waiting for it.

Fortunately, the apartment owners agreed. So now there's a fund quietly growing in the background for a roof that doesn't need replacing today.

And I've often wondered what would have happened if they hadn't agreed. Because saying no would have been completely understandable.

"The roof isn't leaking over my apartment."

"Why should I pay for something we're not doing?"

"We'll deal with it when the time comes."

"I might not even own this apartment in ten years."

And perhaps the most compelling:

"Why should I put real money into an account today for a problem that doesn't exist yet?"

All perfectly reasonable. Until the day the problem does exist. Then you don't have ten years. You have four apartment owners getting water through their ceilings, contractors telling you the patchwork is over, and an invoice for a complete roof replacement. And somebody has to find the money. Now.

That's what I find interesting about this story from an investment perspective. We spend enormous amounts of time discussing whether something is a good investment. We spend much less time discussing when an expense becomes real.

Is the roof replacement an expense ten years from now? Or is it an expense today that simply isn't payable for another ten years?

Those are two very different ways of looking at exactly the same roof. The first says: "We'll cross that bridge when we come to it." The second says: "We're already walking toward the bridge."

There's also an interesting problem for whoever has to manage the property. Nobody particularly likes the person proposing the reserve fund. He's asking you for real money today and giving you absolutely nothing today. No new roof. No nicer lobby. No higher rent. No visible improvement whatsoever. He's selling you the absence of a financial emergency ten years from now.

That's a remarkably difficult product to sell. And yet that's essentially what reserves are. The roof. HVAC. Plumbing. Parking lot. Elevator. Whatever your particular property contains that is quietly getting older while everybody collects rent.

Eventually something will need replacing. The unknowns are "when" and "how much".

This building got lucky. The owners agreed. But what if they hadn't?

And there's another complication for investors that makes the answer less obvious than it first appears. What if you're going to sell the property in five years? Should today's cash flow pay for a roof somebody else may eventually own? Should you sacrifice part of today's return to prepare for an expense that might not happen during your ownership? Or is that simply part of owning an asset - the building is consuming that roof a little bit every year, whether you're writing a check for it or not?

I don't think the interesting answer is simply "always have reserves." The interesting question is: how much of tomorrow's predictable trouble should today's investment be paying for?

Because either way, somebody eventually pays for the roof. The choice is whether ten years of you pay a little at a time - or one very unlucky version of you gets the entire bill.

(When you know a major capital expense is coming years down the road, do you start charging today's investment for it now - or keep today's returns and deal with the expense when it actually arrives?)