“Learn to see in another's calamity the ills which you should avoid.” Publilius Syrus
For investors with a dividend strategy, dividend cuts are skunks at the party, stinking up everything from current income and future income growth, to total return, reinvestment risk and future stock-picking confidence. Calamity indeed.
Not long ago, SA initiated a superb forum, moderated by David Van Knapp, inviting readers to air their choices for the next dividend skunks. It is a firing line of informed opinions, practical experience and good old-fashioned arguing. Recently, David Fish began shipping the firing line more ammo: dividend growers that delayed their hikes beyond the normal anniversary date.
Great stuff, that’s for sure.
And it aroused my curiosity. What might hard numbers tell us about predicting dividend cuts? What warning signs are best, and how good are they? To find out, I turned to a couple of research studies and some historical dividend-cut calculations.
Let’s look at the numbers then kick around some ideas on what to do about them.
In the first study, researchers at Charles Schwab examined several factors, including payout ratio, debt-to-equity ratio, return on equity, and five-year sales growth, for the largest 3,200 U.S. stocks. The research covered 1990 through Q1 2009.
The best predictor? Return on equity. Stocks in the lowest ROE quintile (i.e. bottom 20%) were twice as likely to cut dividends over the following 12 months, with about 14% of them making a chop. Conversely, stocks with average to above average ROEs were the most likely dividend raisers over the ensuing year.
Surprising? I think so, but perhaps not so much after a second look.
My guess, ROE might act as a simple proxy, with high numbers pegging business models and managements able to endure tough times. These top-tier companies’ dividend policies reflect reasoned confidence that better times will come. Meanwhile, bottom quintile ROEs are attached to bottom-of-the-barrel businesses, just scraping along.
An anecdote: last year just one Dividend Aristocrat cut its dividend, wounded food retailer Supervalu (SVU). Its ROE the preceding year was negative 111%; its TTM earnings payout at the time of the cut a seemingly fine 37%.
Before moving to the second study, which researched more factors and cranked more stat power, let’s pause for some other numbers that help inform dividend cut predictions.
Both SVU and Schwab show the relative rarity of dividend cuts in normal years. Of forty-two 2010 Dividend Aristocrats, only SVU cut. Outside the dregs of the bottom-clinging ROE quintile, an annualized 7% of Schwab’s dividend payers made cuts during that two-decade study.
Only four S&P 500 (SPY) companies cut their dividend in 2010 and just twelve in 2007. That’s out of 300+ dividend payers in the index. And finally, of 200+ Dividend Achievers last year, just two cut dividends.
What to conclude from this? Predicting dividend cuts can be tough simply because, except for times like the 2008-2009 meltdown, there aren’t many of them. As stat geeks might say, cuts have a low ‘baseline probability.’ This can create lots of false alarms (i.e. ‘false positives’) as well as surprisingly low prediction power, even with warning signs (ROE or whatever) taken into account.
Look at it this way. If a normal occurrence rate is 5% and prediction methods increase that a whopping 10x, you reach the 50-50 level of a coin flip. Consider Schwab’s study, where ROE moved the needle to 14% from 7%.
The second study, conducted at Brandeis University, boosted prediction success by adding more factors and using heavy-duty statistical techniques (fancy stuff called logit analysis and least-squares regression). The research studied stocks in the Compustat database over the 40 years 1965-2004. Regrettably it covered only non-financial companies (another reason to hate banks) but the results seem sensible enough to consider across a broad range of sectors.
The study uncovered several skunk warning signs. Stocks that stink on these are choice candidates to cut dividends within a year.
- Low return on assets (the study didn’t look at ROE)
- Low sales growth
- Low cash holdings
- High debt
- High price-to-book ratio
- High capital expenditures
- No dividend increase
- Poor stock performance, pushing the yield higher
Taken together, these factors seem to identify firms with weak operating results, poor financial flexibility and competing cash needs. The market recognizes their troubles, bidding the stock down. Also note these factors can vary by industry, so the research benchmarked for that.
About two-thirds of companies loaded with these skunk factors cut their dividends within a year, one-third did not. The key difference between them? Those that did not cut showed a sharp upturn in sales, a rebound in ROA and a reduction in capex.
But highlighting the difficulty of predicting cuts, a number of companies in the study whacked their dividends despite minimal signs of distress in their numbers. Arguably, this is what happened to HCBK when bank regulators nailed it with new balance sheet restrictions.
So what might investors do with all this? If your usual dividend-cut radar starts pinging, sniff the skunk factors along with your other detecting, and take a couple of extra steps as well.
First, weigh those operational and financial numbers to make an overall assessment about the business. Include internal return rates like ROE, ROA, or ROIC which, intuitively or not, seem to make a difference.
Second, in the real world persistently high payout ratios and adequate cash flow matter, so pay attention to them, research aside. It’s likely that payout ratios didn’t generate good correlations because companies with low ratios make cuts, while great cash flow generators, especially troughing cyclicals, don’t cut despite high earnings payouts. Microchip Technologies (MCHP) and Paychex (PAYX) provide recent examples of how cash flow can cover dividends when cyclical earnings tumble.
On a different note, go outside the numbers to check companies’ dividend announcements and earnings transcripts. In addition to PAYX, Meridian Bioscience (VIVO), for example, made clear it policy, payout capacity and plan to maintain dividends. In the 2008-2009 crisis, such pronouncements weren’t worth much. But when under 5% of quality companies are cutting, it’s a different story.
Expect to make some mistakes. The numbers say predicting dividend cuts can be a tough guessing game. Which also means it makes sense to find methods that boost your chances of smelling out the next skunk.