For some reason, I can’t bring myself to shake the old notion that “cash is king.” I get that it hasn’t been the hottest concept on the Street lately, with interest rates as low as they are. But with more than the usual level of uncertainty today (the future of interest rates, the presidential election and what it will mean for a host of issues about which businesses and investors care) not to mention persistent frustration over the slow and not-as-widely-shared-as-we’d-like nature of economic progress, I finally yield to the temptation to develop a cash-cow stock-selection model.

This Should Work

The traditional benefits of hefty cash flows remain in place. This gives business the wherewithal to cope with proverbial and probably inevitable rainy days. It gives companies the wherewithal to grow their businesses without undue (or often any) reliance on debt and debt-related risk. In other words, cash helps companies play better defense. It helps them play better offense. What’s not to like.

This also works when I go Ivory Tower and reason, as I often do, from the traditional dividend discount model, which I extend by substituting earnings for dividends. The formula for an ideal PE is 1 divided by the difference between the required rate of return and the expected growth rate. Cash reduces risk, which in turn, reduces the required rate of return. So all else being equal, increases in cash generation push PEs upward.

Hence practical and theoretical considerations both point to the same conclusion. How good is that!

Following through on basic logic, I expect a backtest of my cash-cow model to show decent results in the past. I also expect results to be at least as good or probably better relative to the market over the past 12 months given that the increased presence of cash should, I assume, soothe the interest-rate and other fears that became troublesome of late.

Surprise, Surprise

I created a Portfolio123 screen that defined excess cash flow as cash from operations (the accrual-free item reported on the Statement of Cash Flows) minus capital spending minus dividends. I focus on the S&P 500 companies excluding financial and biotech stocks because their cash flows often don’t fit this basic mold. I also exclude tech, which normally a more cash heavy sector due to the greater product-cycle needs. From among the “basic” blue-chips I consider, I identify firms that have had positive excess cash flow in each of the past five 12-month periods, those that ranked in the top half in terms of five-year average excess cash (scaled by sales) and five-year excess cash flow growth.

Reviewing passing companies, it is clear the screen successfully picked up the kinds of companies I hoped to see, usually about 50 of them.

That’s too many for the typical individual portfolio, so I used ranking criteria to sort and work my way down to a top 20. And that’s where things when whacky.

Long-term tests stretching back to 1/2/99 had results that were pretty much as expected, decent. But tests for the latest 12 months were surprisingly bad. On one level, I’ve seen this sort of thing before. The last 12 months have been tough and lots of once-good ideas haven’t been working lately. (See, e.g., value, oil.) The problem here is that all the logic I could muster suggested this model should have been stronger relative to the market during the past year.

Is King Cash being impeached?

The Answer

The problem, as it turned out, wasn’t so much that King Cash has been deposed but instead, that it replaced its Prime Minister.

Tables 1 and 2 show longer term (1/2/99 – 4/1/15) and recent (4/1/15 – 4/1/16) tests using a variety of multifactor ranking systems with which I work regularly, as well as a few more focused metrics. (In all cases, I use 3-month refresh periods and assume 0.25% per transaction price slippage):

Table 1 – Test Period 1/2/99 – 4/1/15

Table 2 – Test Period 4/1/16 – 4/1/15

As to the longer-term tests (Table 1), the results are in line with my general expectations including the fact that naïve price momentum would be the least effective way to pick and choose among potential cash-cow stocks.

Value and Quality, the approaches that seemed most theoretically sensible, went off the rails in the latest year. Sentiment, too, diminished. Oddly, though, Momentum, which also softened, was closer to its long-term trend.

Growth is interesting. In theory, historical growth rates upon which the ranking system is based provide some, albeit imperfect, basis for making assumption about the future. And a cash-cow company that seems able to generate growth strikes me as one that might hold promise as an investment. The growth-based model weakened in the past year, but suffered the least amount of diminution relative to the other general styles.

Now, here’s where it gets interesting.

Logic also suggests that a longer-term track record of growth ought to be more persuasive than growth over a shorter period, which is more prone to having been influenced by unsustainable factors. Yet when I drill down within growth, I saw the opposite. Use of the five-year growth rate produced the weakest performance relative to other growth rates. And actually, use of lesser track records accumulated over just a year or a quarter resulted in continuation of what we saw for most strategies from 1/2/99 through 4/1/15.

Choosing cash-cow stocks based on the Sales Growth TTM and Sales Growth PYQ metrics were the only approaches that delivered on my initial expectations (better performance relative to the market during this especially challenging environment).

What It Means

The only metrics that helped us identify worthy excess cash generators during the past year were the most near-sighted myopic approaches I tested.

That reflects the impact of fear. Fundamental common sense is all well and good when things are normal. But when the market is feeling stress, the impact of solid fundamentals can be drowned out by lack of trust in the future. Hence, Mr. Market is, in essence, telling us: “I don’t trust general growth, quality or whatever now because that’s based on data from the past and I have no confidence it will persist going forward.”

Strictly speaking, data for Sales Growth TTM and Sales Growth PYQ also represent the possibly unsustainable future. But Mr. Market is willing to give metrics like these a pass, most likely because of their aura of immediacy. “Who knows if the growth will last well into the future? It’s happening right now. That’s not guru sanctioned but what the heck, I’ll take it, because I’m scared to death about what may be around the corner.”

We get a bit of a confirmation of this through use of the latest year’s Momentum and Sentiment ranks. These indirect indicators suggesting that good things may be happening delivered less than the more direct Sales-based measures. But they held up far more effectively relative to what they delivered in the longer test, and in the latest-year test, even they managed to wallop the more supposedly substantive styles.

Picking Stocks

Looking only at the tests, it’s a no-brainer: Use my excess cash flow screen and pick the top 20 stocks as per as sort based on the rate of trailing 12 month sales growth.

The temptation to do that is strong. But I have to remember the boilerplate “past performance does not assure future outcomes warning,” not only because it’s required but also because it makes sense and should be heeded even if I would be allowed to ignore it.

I’m going to read the message Mr. Market is telling me; that I should pay attention to immediacy. That would have succeeded in the recent past. It would also have been good over the longer term. So I will use Sales Growth TTM, which seems to be an approach for all seasons. Realistically, though, knowing that it’s humanly impossible to completely banish the benefit of hindsight in developing strategies no matter how hard we try, I want to hedge reliance on the test per se with the addition of some theory. I’m also going to work with the Quality ranking system, the most fundamentally thorough one I have. I do that by running each version of the model separately and settling only on stocks that pass both models.

Table 3 shows the test results of this dual approach.

Table 3

Table 4 lists the stocks that currently pass muster.

Table 4

Since I’m now requiring a stock to be picked by each of two models, I’m not surprised to see fewer than the 20 positions we’d be seeing if we looked at just one strategy. Given the special “situation nature” of this excess-cash-generator approach, I’m fine with that. That said, I’d need to fine-tune more if I decide to develop this into a full-fledged automatically refreshed Smart-Alpha model.