The great Warren Buffett hasn't done too badly in insurance stocks. Buffett once owned a bank that was possibly the best-run bank in the country, with a 2.3% return on assets in 1979 (by my rough calculations, that'd put it in the top 2.5% of banks today). However, the Bank Holding Company Act of 1969 required that Buffett choose between insurance and banking. He chose insurance. Yes, insurance, stodgy old insurance, is perhaps one of the greatest creators of wealth in the world. Here's why.

Float like a butterfly.
Float is what happens when policyholders pay insurers up front and don't expect their money back until something bad (like a car accident) happens. Meanwhile, the insurer can invest the money and pocket the returns. Keep in mind, the insurer gets to use other people's money (OPM) to invest for itself.

In 2005, Berkshire Hathaway (NYSE:BRK-A) had $49 billion in float. If it invested this in 4.97% one-year Treasury Notes, it could make about $2.4 billion without taking any risk whatsoever. Cincinnati Financial (NYSE:CINF) used its float to buy shares in Fifth Third Bancorp's stock and turned a $283 million investment into $2.7 billion as of the end of 2005 -- a gain that directly increases the intrinsic value of Cincinnati's stock.

The ability to make money using OPM money is the hallmark of a great business. Banks, mutual funds, hedge funds, private equity funds, and bookies all use other people's money, but what makes insurance superior to all these other OPM users is the fact that insurers can have a positive cost of capital.

Sting like a bee!
In a nutshell, the combined ratio indicates how much an insurer pays out in premiums. A combined ratio under 100% indicates that the insurer is making an underwriting profit. In this case, not only does the insurer get to use the premium to make money for itself, it also doesn't have to pay back the full amount it "borrowed"! It's as if someone came up to you and said, hey, here's a billion dollars, you can do whatever you want with it, as long as you give me $990 million back. Pretty sweet deal, right?

Banks also use OPM. Customers deposit funds at banks, and the banks loan those funds out and collect interest, which it pockets. The key difference is that depositors, in exchange for allowing the bank to use their money, expect to receive interest payments (the insolence!), so banks have to pay them 2%-5%. In other words, the bank has a negative cost of capital. On the surface, mutual funds have it a little bit better. Investors give the fund their money, and the mutual fund gets to keep about 1% of assets -- a positive cost of funds right? Wrong. Whereas banks and insurers keep the profits from reinvestment, all of the reinvestment profits that mutual funds earn belong to the investor, not the mutual fund manager. Not only that, but investors expect mutual funds to earn reasonable rates of return. Otherwise, they yank back their money. This is an implied cost of funds.

In a nutshell, to a well-run insurer, float is an awesome weapon. It can be used to internally fund the business. It can be reinvested for profit. It often generates its own income. Best of all, the insurer keeps all the profit for itself.

Now the bad news
If, by now, you're convinced of the total awesomeness of the insurance industry, it's time to add some caveats. The downside is that insurance involves taking risks. In 2004, Montpelier Re's (NYSE:MRH) combined ratio was 77.8%, meaning the company retained about $0.22 for every dollar it took in as premiums. This was incredibly impressive given that 2004 was, at the time, the worst year ever in property and casualty insurance history. In 2005, which was much worse than 2004, Montpelier Re's combined ratio was 200.7%, meaning it paid out about $2 for every $1 it took in. Needless to say, this resulted in huge losses and the stock took a nose dive. Although reinsurance tends to have more risk than more predictable types of insurance, such as auto insurance, my point is that year-to-year results for insurers can fluctuate, on account of the inherent volatility of the business.

Investing in insurance
I haven't plunged head-first into the insurance industry yet. No, I'm waiting for two things: who and when. "Who" refers to picking the right insurers -- the ones who have low expense ratios, disciplined underwriters (and thus a low combined ratio), and a knack for investing float profitably. The insurers I've come to admire the most are Progressive (NYSE:PGR), Markel (NYSE:MKL), White Mountains (NYSE:WTM), and Berkshire Hathaway. I also like Zenith (NYSE ZNT), Torchmark, OldRepublic, Cincinnati Financial, and Mercury General (NYSE:MCY).

"When" refers to timing. I prefer to buy insurance companies when blood is in the street. Why? Because I know if a disciplined underwriter like Progressive is writing policies at a 100% combined ratio (i.e., breaking even on underwriting), then others are probably losing money and will have to cut capacity or go out of business -- meaning more future profits for Progressive. This also helps me buy at low valuations (preferably at a low price to tangible book ratio). So far, I've only gotten the chance to buy Montpelier Re after Katrina shellacked the industry, but if I'm patient, I'll be ready at the next downturn.

More insurance-related Foolishness:

Mercury General and Montpelier Re are Income Investor recommendations. Berkshire Hathaway is an Inside Value recommendation.

Fool contributor Emil Lee is an analyst and a disciple of value investing. He owns shares in Montpelier Re. The Motley Fool has a disclosure policy. Emil appreciates comments, concerns, and complaints.