Over the years all of us have had many teachers, professors, mentors and so forth but only a few who come up with sound bites so memorable, they remain in your mind decades later. For me, one such person is Richard Farrell of Brooklyn Law School (law being my first career) who often responded to what his civil practice students thought to be profound insights by saying "That sounds good if you say it fast enough." As we go through the process of (hopefully) solving today's financial crises and digest the avalanche of ideas and opinions that will undoubtedly continue to come our way, I suspect Professor Farrell will come to mind quite often, as was the case in the past few days, when I read a couple of articles on something known as the accrual ratio, and the "cash is king" notion from which it springs.
Why use accruals
Essentially, accrual accounting describes situations where reported numbers reflect management estimates, rather than an actual count of cash that comes in and goes out. During economic and/or market down-cycles, it becomes popular to criticize this practice for not giving a true picture of corporate performance and to suggest that the best measure is free cash flow, something that supposedly tells us how much the company really has available to pay debt, to pay dividends, to invest back into the existing business or into new opportunities, and so forth. Another often-cited benefit of free cash flow is that it's less susceptible to management manipulation, the theory being that if you allow management to make estimates, that privilege may be (or, according to the harshest critics, will be) abused.
There's no doubt abuse is possible. Actually, though, regardless of accounting rules or situations, bad managers can always find ways to do bad things. But we ought not let this blind us to the fact that accruals provide very valuable information that ought not be lightly discarded. According to Graham & Dodd's Security Analysis (Cottle, Murray, Block, 5th Ed., page 143):
It would be nice if all companies liquidated to an all-cash position at the end of each year and satisfied all their liabilities — there would be few quarrels about the amount of profits. Unfortunately, many cash cycles are still going on at the end of the year. Some of them are quite short cycles and others extend over many years. Thus, accrual accounting is designed to record as well as possible the performance of the company, using various conventions that capture the part of each cash cycle that was completed during the period.
The classic example is depreciation. This is a estimate prepared by management based on accounting guidelines as to how much of the physical asset base is used up each year. Nobody pretends it's dollar-for-dollar accurate. But many cash-is-king advocates suggest it's less reasonable than to simply look at actual capital expenditures.
That sounds good if you say it fast enough. But when we slow down, we see that capital expenditures can be volatile from year to year, as major projects ramp up or wind down.
Assume a company spends $100 million for a factory that will last five years. Assume too that goods produced by that facility generate $80 million in annual revenue and $70 million in recurring annual cost.
Free cash flow in year one is minus $90 million, a $10 million profit from operations minus $100 million in capital expenditures. That's a huge deficit.
In years two through five, annual free cash flow is $10 million, which is the profit from operations (assume it's all cash).
One who stumbles upon the company in year three and starts analyzing will see annual free cash flow of $10 million. But is that really available for dividends, share repurchases and so forth? In just two years, the firm will need to invest heavily in physical plant in order to stay in business, so we'd have cause for concern if it is using its free cash flow for dividends or share repurchases. Actually, if we depreciate the plant evenly over it's five-year useful life (accrue a $20 million annual depreciation charge), we'd see the firm was losing $10 million a year and absent some sort of structural change, would not likely be viable.
The accrual ratio
I can tell times are tough because twice in the past two months I've seen articles advocating something called the accrual ratio, which is net income minus free cash flow (cash form operations minus capital spending minus dividends) with the result divided by total assets. The smaller the number (the smaller the accruals), the better it is.
Figure 1 presents the results of Portfolio123.com backtest of a one-factor ranking system based on this ratio.
Figure 1: Accrual Ratio, 3/31/01 - 9/17/08
The leftmost red bar depicts the annualized performance of the S&P 500. The next dark blue bar depicts the performance of the stocks ranking in the bottom five percent. The rightmost bar shows the annualized performance of the best five percent, etc. Note too that this and all other backtest results are for non-financial companies only which have share prices above 3 and market capitalizations of at least $250 million. Bulletin board stocks are excluded across the board.
The result is somewhat respectable. The lackluster showing of the top-ranked stocks is noteworthy, but those who create investment models have often seen situations where the extremes don't conform and will address it with follow-up work. But it is a reasonable start.
Even so, I remain unimpressed. Why divide by total assets? It looks to me like this isn't so much a statement of earnings quality, as its proponents claim, but a variant on the return-on-assets theme. Instead of dividing net income by assets, we're in effect dividing the surplus of free cash flow over net income by assets (assuming we flip the subtraction so higher numbers would be considered good).
That's a fine notion. But Figure 2 suggests plain-vanilla return on assets works better.
Figure 2: Trailing 12 Month Return on Assets, 3/31/01 - 9/17/08
Speaking more slowly
If we really want to measure the impact of accruals, we should divide the free cash flow-net income gap by net income rather than assets. That way we get the percent of net income that reflects accruals.
I also disagree with subtracting dividends from cash from operations. A more apples-to-apples comparison would be free cash flow versus retained earnings (net income minus dividends) rather than net income. Actually, though, I prefer an alternative: to look at accrual and cash results without subtracting dividends from either. That way, we're measuring resources presumably available for dividends, debt repayment, share buybacks or business investment for all companies regardless of dividend policy.
Here's my revised accrual ratio:
(Cash from operations minus capital spending - Net Income)/Net income
Figure 3 shows the results of a backtest of a single-factor model based on this approach.
Figure 3: Adjusted Accrual Ratio, 3/31/01 - 9/17/08
That is not good.
Yes, I made some adjustments. But in deciding whether that was fair, consider that by abandoning the original ratio, I moved away from an ok-but-not-great return-on-assets variant and toward something that really seeks to measure the degree to which reported net income reflects accruals. And I now look at all companies equally in terms of their ability to pay dividends (among other things) should they so choose, rather than muddying the waters based on decisions of some companies to pay no dividends, some companies to pay small dividends and some companies to pay large dividends.
A different approach to the core problem
Accounting fraud and abuse are definitely things to worry about. But realistically, these are more in the nature of man-bites-dog rather than dog-bites-man, which is why they are featured by broadcast and print journalists trained to emphasize the former and ignore the latter. And to the extent fraud is relevant, bear in mind there are a lot of smart people out there with finance and accounting degrees at all points on the moral compass. As to those inclined toward wrongdoing, it'll take much more than the accrual ratio to blow their often-sophisticated covers.
That's not to say we should simply accept traditional net income as sacrosanct. In fact, making adjustments is a core activity in security analysis. But let's consider why, in the normal workaday (i.e. ex criminality) world we make these adjustments.
Start with the notion that financial statements tell us about the past, yet investment performance is based on the future. But we take historical data seriously because, according to Graham & Dodd, "a tendency toward an underlying continuity in business affairs makes the financial record the logical point of departure for any future projection" (Id. at 150).
When it comes to probable continuity, not all elements of the income statement are equal. Many items are specifically labeled as being unusual, sometimes by management and at other times by the major data vendors. Step one in creating an earnings estimate is to re-cast the income statement in a way that isolates that which is likely to be ongoing from that which is not. (At least that was how it was before one could simply get official company guidance.)
When I look at an income statement, rather than worrying so much about accruals, I try to give myself as much visibility as I can get on those trends that are most likely to persist. These are the ones that relate to revenues, cost of sales, selling general and administrative expense, and depreciation. I will also look at interest expense, but there's value in also looking at the pre-interest number, which allows me to compare different companies which may have chosen different ways to structure their balance sheets.
Many may recognize EBIT (earnings before interest and taxes) here. That is the general idea. Bear in mind, though, that EBIT numbers calculated by modern data vendors tend to use as a starting point "operating profit" as reported by companies, a number that includes many kinds of items that are not likely to persist from one period to the next and other items that may persist but relate to fringe investment activities rather than the core business. Let's call my number C-EBIT (Continuing EBIT).
I suggest the larger the percent of reported operating profit that comes from continuing C-EBIT, the better. This doesn't mean I ignore the other items. But I see C-EBIT as an important starting point.
Creating a formula that's analogous to the accrual ratio, we have C-EBIT minus reported operating profit, with the result divided by operating profit.
(Core-EBIT - reported operating profit)/reported operating profit
Figure 4 shows the backtested performance of a model based on my C-EBIT ratio.
Figure 4: Continuing-EBIT Ratio, 3/31/01 - 9/17/08
Such a single-factor model wouldn't by any means be the be-all and end-all of stock selection. But clearly, it is much more useful than the accrual ratio. The C-EBIT ratio at least showed some potential to steer us away from a large group of conspicuous laggards. In other words, the market didn't really reward pristine income statements, but it did seem to punish companies whose reported operating profits relied most heavily on things other than plain-vanilla business line items.
No more kings
It does seem that the expression "cash is king" sounds best when said quickly.
I'm not sure I'd feel comfortable identifying any other single factor as an alternative. But to the extent we are going to dig deeper into the numbers reported on the income statement, it looks like the kingdom would benefit from efforts to distinguish between items that represent ongoing trends in the core business versus items that might be classified as "miscellaneous," "unusual," "other," and so forth.