In Part 1, I analyzed 56 stocks that cut their dividend and observed that for 75% of them, their stock price underperformed the S&P 500 (NYSEARCA:SPY) by 20% (2000 basis points) within 12 months of the dividend cut announcement. Furthermore, selling at the signal would have saved investors an average of 22%, as opposed to waiting until the announcement. I hypothesized that this performance gap could be used to anticipate the dividend cut and exit the stock before the price declined further. An important question that was unanswered in Part 1 was how often this signal is wrong. That is, how often does a stock have a -20% performance gap relative to the S&P 500, yet continue to increase its dividend?
To address this question, I performed the same analysis on the stocks in my DCC Model Portfolios (DA+, Small-Cap, Income-Growth), all of which have consistently raised their dividends annually for 7+ years, most over 15 years. These stocks represent all sectors and include large and small caps; see the list at end of article.
Because these stocks have not cut their dividend, I didn’t have a specific time reference (i.e. a cut date) for my chart analysis. Therefore, I looked at one-year charts starting from five years ago (Oct 10, 2006), and recorded the date if I observed a 4-week -20% gap. In some cases where I could see larger price swings, I altered the starting month to capture that change, since an investor would likely be performing this test on a monthly or quarterly basis. I admit that this process is not the most precise, however, I consider the results good enough for the purposes of this discussion, as I was purposely trying to find gaps. Over half of the stocks tracked the S&P 500 very closely; it’s the performance of the other group that I will analyze more thoroughly.
Of the 70 stocks in the sample group, 39 (56%) did not trigger the -20% performance gap rule; the model correctly identified that these stocks would not cut their dividend. I was hoping for a higher percentage, as a coin toss has a 50% chance, but considering the simplicity of this indicator, I’m not too surprised. The model was very accurate (82%) for the Utilities, Info Tech, and Consumer Staples sectors. The more interesting results came from the group that the model incorrectly predicted would cut dividends.
Incorrect, from a certain point of view…
The model incorrectly classified stocks 44% of the time, however, I was curious about the performance of those stocks despite the lack of a dividend cut. For the 31 incorrectly flagged stocks, I recorded the price at the signal date and the lowest stock price within the next 12 months, as well as the SPY’s prices for the same dates. The predictor suggests that the stock will cut its dividend in the next year, so I expected that the price should fall, as the company should be struggling. While these stocks did not cut their dividends, here are some interesting observations about the stocks in this group.
- On average, their prices fell 22% between the signal date and their lowest respective point over the next year. Interestingly, this is the same average drop for the group that actually cut their dividends.
- On average, the S&P 500 (SPY) fell just 13%, so this group underperformed the market for the measured time periods.
- 22 of 31 (71%) experienced price declines of over 10% within the next 12 months; only 2 had no decline. This suggests that most of the time, DG investors could avoid additional losses and potentially reenter the stock at a lower price if the dividend appears safe.
- The Financial sector had the highest percentage of mislabeled stocks (9 of 11), however most of them declined another 20%+ after the signal. Note: Given that these data mostly come from the recent meltdown, it may overstate losses for financials in the future.
- If the stocks in this group were not sold and instead held to the present time, their average price change would be +1%, though excluding one outlier (HGIC, which was recently acquired for 100% premium), the average would be -3%. While it is encouraging that as a group, the losses were mostly recovered over time, 19 of 31 (61%) stocks are still negative on a price basis, and the investor had to hold these stocks around 3 years on average to just breakeven. I would like to think that investors deserve, and can do, better!
- 17 of 31 (55%) stocks only issued 1-cent dividend increases over the last few years, for an average growth rate of 3.8%. 9 of 31 (29%) are overdue to increase their dividend, a warning sign to DG investors. So, while the model incorrectly flagged these stocks for a dividend cut in the next 12 months, the group does not exhibit a particularly high dividend growth rate, and many are late (past 4 quarters) for their next increase. The signal may have just been a little early. Switching these stocks for better options may not be such a bad idea.
While technically the model incorrectly labeled these 31 stocks for dividend cuts, from a performance standpoint, selling these stocks would generally have been the right move. Nearly 3/4 of these stocks experienced 10%+ price drops within a year of the signal, the group underperformed the S&P 500, over half had small dividend increases, and 29% are late for their next increase. Assuming an investor can find a replacement stock with the same yield, the dividend stream is preserved and hopefully a higher quality stock has joined the portfolio. Yes, a few of these stocks appreciated, but since capital gains are not the priority, I believe DG investors should be more concerned with preserving principal and the dividend growth rate. As the saying goes, “better safe than sorry.”
Combining the results from Part 1 (known cuts) and Part 2 (known non-cuts) yields the following summary table with a sample size of 126 stocks. The green sections indicate correct predictions, the yellow box represents when the model predicted a cut but no cut occurred, and the red box highlights undetected dividend cuts.
An Action Plan for DG Investors
- Red Box: If a DG investors is simply selling stocks after a dividend cut is announced, then effectively, his/her div-cut stocks are in the red box and will experience whatever price losses occur prior to the cut. Based on the sample data, the -20% signal would have detected those cuts 75% of the time and saved the investor 22% on average. I'd be willing to go with those stats.
- Yellow Box: Many DG investors say they don’t mind price declines as long as the dividend continues to rise. In this case, they would hold on to all of the No-Div-Cut stocks and collect the dividend, often with small increases. While the overall model incorrectly predicted dividend cuts in 24.6% of the cases, it mostly identified stocks that continued to fall over 10% in price, had meager and sometimes overdue dividend increases, and had no price gain on average since the signal date. One option is to sell and then buy the stock back at a lower price point if you really want that stock. The data suggest that in most cases, you would get this opportunity.
Option #2 is to switch to another stock with a comparable yield that hasn’t experienced the -20% gap. This would maintain the dividend stream, hopefully provide better dividend increases, and avoid the underperformance of the yellow-box group. As long as the dividend stream is preserved, the DG investor’s primary goal is met. Given the observational data, wouldn’t it be worth the risk to try another stock to avoid the underperformance and low dividend growth rate? Sort of like being on “Let’s Make a Deal”…I would opt for Door #2 after seeing the collective traits and results of the yellow box group.
I believe this signal is a worthy indicator of underperformance (price and/or dividend) for a consistent DG stock. 80% of the time, it accurately predicted a no-cut, a dividend cut, or a continued price decline of over 10%. That’s a bit better than a coin toss! Applying additional fundamental analyses (payout ratio, earnings & cash flow analysis, business outlook, debt, etc) would help investors to make more informed decisions if they don’t want to just trust a single metric. I always recommend gathering and evaluating more data. Time will tell if this simple price change indicator is as accurate and helpful to investors’ total return performance as it appears based on the sample group analyses. The data suggests that price can convey useful information to investors, so DG investors should not ignore changes in price, and hopefully it will help them to avoid dividend cuts and capital losses.
New Model Portfolio
To test this exit signal, I will create a variation of the DCC-IncomeGrowth portfolio effective October 14, which removes stocks that exhibit a -20% performance gap within the last 12 months. This portfolio will be more actively managed during the year to remove stocks that develop the signal. I would expect this rule to have more impact during down markets, as few companies typically cut their dividend in up markets and earnings usually don’t decline. Had I applied this process when I created the DCC-IncomeGrowth portfolio in August, Avon (NYSE:AVP) and Meredith (NYSE:MDP) would not have been added because they triggered the signal earlier in 2011. Both proceeded to fall another 20%+ after the signal, but continued to raise dividends, though AVP raised its dividend by only 1 cent. The new portfolio will replace these two stocks with Wal-Mart (NYSE:WMT) and Genuine Parts Co (NYSE:GPC) respectively.
No-Cut Dividend Sample Group Stock List:
MDP, LEG, GPC, LOW, MCD, CLX, SYY, PG, AVP, UL, KMB, MO, UVV, PEP, CVX, COP, SXL, EPD, KMP, MUR, XOM, CFR, HGIC, MCY, NHI, ORRF, OVBC, UBSI, OTCQX:HARL, CINF, AFL, CB, MDT, JNJ, ABT, NVS, OMI, SPAN, GD, MGRC, EMR, MMM, SWK, ABM, MPR, WSO, PBI, LLTC, MCHP, ADP, DBD, INTC, RPM, APD, BMS, PPG, SON, NUE, T, CTL, VZ, ATNI, NST, NEE, UNS, BKH, SCG, ED, MSEX, VVC