It's a logo. Photographer: Tim Boyle/Bloomberg News.

I fell a bit down the rabbit hole of Nuveen Investments this morning, spurred by Felix Salmon's puzzling over the cash flows. The puzzle is this:

• Madison Dearborn and others bought Nuveen in 2007 for a valuation of around \$5.75 billion, with private equity kicking in around \$2.7 billion of that value and debt contributing around \$3.05 billion.
• Nuveen paid roughly no dividends between 2007 and now, but incurred an extra \$1.4-ish billion of debt, for a total of \$4.5 billion of debt now.
• Madison Dearborn (and the others) are selling Nuveen to TIAA-CREF for \$6.25 billion now.
• \$6.25 billion minus \$4.5 billion of debt is \$1.75 billion of proceeds.
• \$1.75 billion is less than \$2.7 billion.
• Yet Madison Dearborn "will have at least broken even."

This particular puzzle seems to be solvable through a series of complicated and not entirely public moves, of which the most important is that Madison Dearborn gets "\$600 million or so" of cash outside of the purchase price, including Nuveen's balance sheet cash and the purchase price for some other side items. So the private-equity investors are getting back around \$2.4 billion of their \$2.7 billion; combine that with an earnout, another side investment and the fact that Madison Dearborn itself bought out a co-investor's stake at a discount during the crisis, and Madison Dearborn seems to have done OK.1

Fine. But Salmon also frames the puzzle in another way, not as a sources and uses puzzle but as a net income puzzle:

The cumulative losses that Madison Dearborn has overseen come to somewhere in the region of \$2.4 billion.

So here’s my back-of-the-envelope math: you buy a company for \$2.7 billion in cash, plus debt which you refinance a few times. While you’re running the company, it loses a total of \$2.4 billion. And then you sell the company for \$1.75 billion in cash. Total going out the door: \$5.1 billion. Total coming in, at exit: \$1.75 billion. Net loss: some \$3.35 billion, give or take.

This is true: From 2008 through 2013, Nuveen had a cumulative net loss of around \$2.3 billion.2 And yet its debt investors have not been haircut, and its equity investors lost ... well, on the order of zero dollars. So how did a company lose \$2.3 billion while the people who own it lost roughly nothing?

The answer is that TIAA-CREF, in buying Nuveen, more or less pretended those losses didn't happen. In other words:3

So TIAA-CREF paid \$2.4 billion for the equity of Nuveen, when Nuveen's financial statements say it was worth \$0.6 billion because of all those losses.

And why did TIAA-CREF pretend that all those losses didn't happen? Because they didn't. The answer is basically this rearrangement:4

That is, of Nuveen's \$2.3 billion of net losses from 2008 through 2013, more than all of them -- \$2.6 billion -- came from a couple of write-downs of intangible assets in 2008 and 2012. Almost \$2 billion of that was in 2008, a rough time to be running an asset management business.5 These were not "real" losses, where cash went out the door. Rather, they were accounting-driven re-valuations of the stuff -- the brand, the customer relationships, the expectations for the asset management business -- that Madison Dearborn had just bought. Madison Dearborn bought near the peak in 2007, at a high valuation corresponding to its optimistic expectations for the business.6 And it rapidly had to write off a chunk of that valuation as it revised its expectations to match, y'know, 2008.

But then things got better. And intangible write-downs only go one way: Generally accepted accounting principles required Nuveen to take \$2 billion in paper losses when the hopes and dreams of 2007 gave way to the sullen despair of 2008, but did not allow it to take any corresponding gains when 2008 gave way to the sunny renewal of 2014. Not as sunny as 2008, I mean, but, y'know, 69 percent of the way there. The only way for Nuveen to write expectations back up to 2014 levels was to be sold to someone else for cash and book the gain. So that's what happened.

Salmon says:

This is worth remembering, when private-equity types (think Mitt Romney) claim that their interests are aligned with the interests of the companies they buy. That certainly doesn’t seem to have been the case here. Nuveen is being sold with about \$1.5 billion more debt than it started with, and with cumulative losses under Madison Dearborn’s ownership of some \$2.4 billion.

But I think there are some other lessons here. For one thing, when you note that those \$2.4 billion of cumulative losses were all paper write-downs of intangible assets, you might think it's kind of nice that they were all incurred while Nuveen was in patient private-equity hands. Public shareholders were cashed out at the top, and then Madison Dearborn patiently sat around until the environment improved enough that it could make its money back.

More generally, this deal is another good reminder that financial statements in the financial industry don't mean anything beyond what they say, and maybe not even that. A bank's balance sheet only gives you one partial view of its assets, and an asset manager's income statement gives you only one partial view of its income. Financial companies make a little money here and there running their business, and then sometimes they are buffeted by larger outside forces that cause billions of dollars in paper losses: \$2 billion for Nuveen in 2008 for the deterioration of the asset-management business generally, say, or \$5.7 billion for Bank of America in the second quarter of 2013 for interest-rate losses in its available-for-sale portfolio. Are these "real" losses? That is sort of a silly question. They are neither real nor fake; they are just one way of looking at the situation.

It is the nature of financial companies that their GAAP net income, their cash flow and the economic value that flows to their owners are three very different things, and that using, say, accounting net income as a substitute for economic value-added will lead only to confusion. That's what makes them fun. Or maybe not fun, exactly, for Madison Dearborn, but better than it could have been.

1 That paragraph is based on Dan Primack's account. Bloomberg News has some information about the co-investor stake, and about the splits among investors.

2 You can get net income for 2009-2013 here, and for 2007-2008 here. My slight differences from Salmon's numbers seem to be mostly about whether you look at net income or attributable income.

3 Dollars in millions. For income numbers, see footnote 2. For shareholders equity in 2007, see here (page 2). For equity, debt and cash in 2013, see here. The \$122 for other changes is a plug to reconcile 2007 equity plus 2008-2013 income to 2013 equity. The \$300 for other goodies being given to equity holders comes from Primack's "combined \$600 million or so" for the balance sheet cash plus the other goodies.

4 For the 2008 impairment, see pages 18-20 of the 2010 financial report (and pages 11-15 of the notes to the financial statements). For the 2012 impairment, see page F-14 of the 2013 Form S-4.

5 As Nuveen said:

Beginning in late 2007, global economic conditions deteriorated to such an extent that we had to adjust our underlying assumptions to take into account the impact the global economic downturn had to prospects for future growth.

6 Just for giggles, here are the projections that Nuveen's board used in selling to Madison Dearborn:

That's \$808 million of projected net income between 2008 and 2010. The actual number was \$300 million if you ignore the intangibles write-downs. It was negative \$1.7 billion if you don't.