In Defense of 'Fuzzy Math'

Sometimes those earnings adjustments make sense and can actually be useful.
At Closing, June 20st
49.89 USD

When dealing with profit figures that don't conform to generally accepted accounting principles, there is one generally accepted principle among the many skeptics in the post-Enron world: Don't trust them.

It's easy to see why. The non-GAAP numbers usually tend to make the company appear healthier than it would be if its bean counters had counted the beans in the official manner. Companies call these figures something like "adjusted" or "pro forma" or "cash" earnings, and the skeptics call them "fantasy numbers" and "earnings before bad stuff" or "fuzzy math."

There is no doubt that earnings of late are fuzzier than the faces at a Brooklyn beard competition. For example, some not-so-fuzzy math by Gadfly's Shira Ovide showed recently that the aggregate profit of tech companies in the S&P 500 was increased last year by $45 billion, or 23 percent, when converting from GAAP to non-GAAP earnings.

The backlash seems to be growing. For example, UBS analyst Steven Milunovich made the case this month that Apple, Rackspace, IBM and Cisco deserve higher price-to-earnings valuations because they are less liberal when it comes to excluding items from earnings. Hedge fund manager Dan Loeb said last year that he was basing short bets on companies that rely too much on dubious accounting.

Different Stories

Non-GAAP earnings figures can lead to wildly different valuations when compared with GAAP numbers

Source: Morgan Stanley

Note: based on last 12 months of earnings.

But you may not want to shave the fuzz off all the numbers just yet. First, it turns out a widening gap between pro-forma earnings and GAAP earnings has simply not been a good tool for timing markets in the past, according to Morgan Stanley strategist Adam Parker and his colleagues. 

The current blowout between GAAP and pro-forma earnings for S&P 500 companies is about 30 percent, according to the Morgan Stanley analysts. That's the widest it has been since the third quarter of 2009. The divergence was even larger in 2001 and 2002. In other words, bear markets were already roaring when the spread between GAAP and fuzzy-math earnings blew out to extreme levels. (Deteriorating earnings, both of the GAAP and non-GAAP flavors, may be the bigger issue if you're in the market for things to worry about. )

In fact, using the adjusted numbers could be the wiser move when picking stocks, according to Parker. That's because many of the items that disappear when the earnings figures are adjusted truly are one-time or nonrecurring issues that won't affect the companies' prospects for future income. Stocks with cheap valuations based on pro-forma numbers tend to outperform those that are cheap based on GAAP numbers, he wrote. 

Still, there are two types of fuzz on the numbers. There are the true one-time or nonrecurring items that are -- for lack of a better phrase -- generally accepted under non-generally accepted accounting principles. (Read that sentence again if you need to -- we'll wait.) The other is the item being passed off as nonrecurring that keeps happening quarter after quarter. Plenty of questionable items are being excluded, and the list is growing. And there are plenty of examples of companies engaging in the fuzziest style of math that proved to be nightmare investments. However, rather than writing off all pro-forma earnings as suspicious, going through the fuzz with a fine-tooth comb is required.  

Morgan Stanley studied the 50 companies that accounted for most of the spread between GAAP and non-GAAP earnings and concluded that half of the adjustments in 2015 were valid, such as impairments and discontinued operations -- largely destruction caused by the plunge in oil prices. The other half were things like costs for stock-based compensation and other items that arguably have no business being excluded from earnings. (The percentage of valid adjustments for technology and health-care stocks in the Nasdaq 100 was much lower.)

Dissecting the Fuzz

Morgan Stanley analyzed the adjustments of the 50 companies that account for most of the spread between GAAP and pro-forma earnings and found that about half were true nonrecurring items

Source: Morgan Stanley

Adjusting for just the valid charges, and ignoring the bogus items, results in valuations that will perform even better than multiples based on either GAAP earnings or the company's own adjusted results, according to Morgan Stanley.  See here:

Warm and Fuzzy

A Morgan Stanley study shows that using GAAP figures in price/earnings multiples to identify stocks to bet long and short does not provide better returns than using pro forma numbers, especially modified pro forma numbers that exclude dubious items

Source: Morgan Stanley

Note: Cumulative monthly returns

Note that these returns are based on a long-short portfolio, where stocks with low price-to-earnings valuations are bought and those with high valuations are sold. The point is not to argue that this approach alone is a market-beating strategy. (It's not --  the S&P 500 returned 1,100 percent from 1985 to 2015, or more than double that when including dividends.)  Rather, it's demonstrating the notion that using GAAP earnings in valuations is not a more effective factor in picking stocks to long and short than using pro-forma earnings (and especially  pro-forma earnings scrubbed to remove items that arguably are not really non-recurring.) 

As such, it's a pretty compelling argument to trim some of the fuzz on the math, but don't shave it off completely.

This column does not necessarily reflect the opinion of Bloomberg LP and its owners.

    To contact the author of this story:
    Michael P. Regan in New York at

    To contact the editor responsible for this story:
    Daniel Niemi at

    Before it's here, it's on the Bloomberg Terminal.