Earlier this year, a government panel released a tome of information and interviews on the financial crisis. One part that caught my attention was the degree to which bank CEOs claim they were unaware of what was going on within their organizations. Frankly, I believe them. Former Citigroup (NYSE: C) CEO Chuck Prince wasn't aware of some mortgage-origination credit lines that he "would not have approved, had he known," according to the report. It wasn't until the bank was already in over its head that Prince says he learned what was really going on in Citigroup's investment banking division as well.

Yet he was the boss. He was in charge of these divisions. Determining risk levels, overseeing assets, approving positions, appointing employees. He was calling the shots.

Is it any wonder Citigroup fell so hard? Too big to fail it was, but too big to manage was its problem. Here you had $2 trillion worth of retail banking, investment banking, wealth management, currency trading, commodity arbitraging, insurance, hedge funding -- God knows what else -- and Prince, a former attorney, running the whole shebang.

Investment great Peter Lynch calls this "diworsification": An unwieldy mess of businesses crammed under one roof managed by one person who doesn't have a clue what's going on.

The list of conglomerates that became too big to manage is long. Tyco International (NYSE: TYC), which did everything from security alarms to suture needles. AIG (NYSE: AIG), which insured your car and traded derivatives on your house. General Motors (NYSE: GM), which learned that it's pretty good at making cars, but terrible at making mortgages, where it lost billions. Even General Electric (NYSE: GE), often hailed as a successful conglomerate, became a cumbersome confusion over the past decade.

If there's a common theme to conglomerates, it's that they're a disaster to manage. Shareholders often learn this the hard way.

But then there's Berkshire Hathaway (NYSE: BRK-B). Berkshire has more than 50 subsidiaries, ranging from underwear to private jets to chocolate. It's the ultimate conglomerate. And if you haven't noticed, it's been somewhat successful.

Why?

Warren Buffett unveiled one reason in his annual letter to shareholders last weekend. Buried on the last page and almost entirely overlooked by media outlets, Buffett included a copy of the two-page memo he sends the CEOs of Berkshire's subsidiaries every two years. The memo starts by reminding mangers that reputation is more important than profits, that they don't need to hide poor results, and they should act immediately when they spot mischief. But it's this passage that's truly meaningful:

Talk to me about what is going on as little or as much as you wish. Each of you does a first-class job of running your operation with your own individual style and you don't need me to help. The only items you need to clear with me are any changes in post-retirement benefits and any unusually large capital expenditures or acquisitions.

That's it. Run your business and let me know what's up. If you feel like it.

This is why Berkshire is successful as a conglomerate: Its management is completely decentralized. Every subsidiary is independently managed from A to Z with essentially zero input from Buffett.

Think of it that way, and Berkshire is only a conglomerate in the technical sense. Rather than a mix of businesses lumped under one roof, it's really a collection of independent businesses owned by one group of shareholders.

This gives Berkshire latitude to buy businesses it never realistically could if Buffett were running day-to-day operations. Take its acquisition of Burlington Northern. Last year, someone asked Buffett's partner Charlie Munger how he and Buffett handle such a massive and complex organization as Burlington. Munger was quick to reply: "We don't have to do one damn thing other than let [Burlington CEO] Matt Rose do whatever the hell he pleases. We don't have to deal with problems."

We don't deal with problems! Chuck Prince had to deal with problems. Rich Wagoner of GM had to deal with problems. Because they were their problems. Every Berkshire subsidiary comes across trouble, but each independent CEO takes 100% responsibility for it. And those CEOs know their own companies inside and out, the opposite of most conglomerate bosses who try to run operations they have no business touching. This leaves all of Buffett and Munger's time to do what they do best -- finding the next great investment.

Don't take my word for it. Asked what Berkshire's key to success was, here's how Munger responded last year:

There are two main reasons Berkshire has succeeded. One is its decentralization of subsidiaries. We've installed decentralization almost to the point of abdication. There are only 28 people at headquarters in Omaha.

The other reason is our extreme centralization of capital deployment. [He and Warren are the only ones who make investing decisions.] Our centralization is just as extreme as our decentralization.

Maybe most importantly, this extreme decentralization does something special to Berkshire that most investors overlook. There's a constant, intense, fear that when Buffett dies, Berkshire will, too. It's understandable -- he's the face of the company. But when you realize how decentralized Berkshire is, and how insignificant a role Buffett plays in the day-to-day operations of Berkshire's investments, those fears look overblown. Long after Buffett's gone, Burlington Northern will be hauling freight. GEICO will be selling insurance. See's will be selling chocolate. Coca-Cola will be paying dividends. These companies won't need Buffett to succeed in the future. Truth is, they don't even use him now.

Check back every Tuesday and Friday for Morgan Housel's columns on finance and economics.