I have been wondering how one might model high-yield dividend investing for a while. One of the challenges of modeling any aspect of investments is to find relevant data. Sales material rarely reveals the truly relevant information. One needs to know something about historical failures of one kind and another.
In this context, I was pleased to run across the information shown in Figure 1. These data span 20 years, from 1996 to 2015. This is shorter than one might prefer, but does span two recessions and two monster bull markets. This suggests that these data represent a reasonable basis for forecasts.
The figure shows the forecast yield, at which one purchased the stocks, and the realized yield, which is what actually occurred after dividend cuts and other negative events. The realized yield drops significantly below the forecast yield beginning at about 7%.
One can see that overall it makes little sense to seek portfolio yields above the level of 7% for common stocks. One might do so for specific companies, or perhaps specific market sectors, if one believes one has good reason. But one must respect that many have failed in the past while reaching for yield.
Figure 1. Forecast yield and realized yield for the period 1996 to 2015. Source
The Monte Carlo Modeling
The model approach used in this article is Monte Carlo simulations, which I introduced and discussed here. If you have let any investment advisor work up an evaluation for you, or used the web tools that are widely available, then you have almost certainly been shown results of a set of Monte Carlo simulations. (This paragraph and some others are reproduced from previous articles, to set context for new readers)
The main question Monte Carlo simulations can help with is to ask what spread of outcomes will be likely for some investments, if the markets behave statistically as they have in the past. This, in particular, can help one understand sequence of returns risks and the distribution of likely outcomes they produce.
For the modeling discussed here, the portfolio has an initial value of $500k and the retiree spends a set percentage of the initial portfolio value each year, in real dollars, for 30 years.
Details Of The Modeling Assumptions
To move from data on total return to a model requires a number of additional assumptions. These are plausible but not certain, and may well be imprecise. My view is that they will successfully capture the relevant trends. Even so, it is only large differences in outcome that will be meaningful here.
The model makes the following assumptions:
The total return of the first three models does not depend upon the amount of dividends or whether they are paid. This will first let us clearly see the impact of dividend payments. But there is evidence that dividend payers do deliver greater returns on average over time. We will incorporate this added aspect by comparing with results from models using the Dividend Aristocrats and equity REITs (eREITs), for which we have specific data.
The company will sustain the yield as the stock price gradually increases. The dividend will mirror the long-term appreciation of the stock price. If this did not happen on average, one would see a lot more variation between realized yield and forecast yield than one does. This will be modeled as a steady increase of the dividend, but in reality, we know that things are messier, especially as one goes up in yield.
The total return will be comparable to that of the S&P 500. If this were not true, then in the long run the various stocks would be bid up or down to make it true. Any residual differences will be at the level of a percent or so. The Dividend Aristocrats are an example.
The variability in the total return is comparable to that of the S&P 500. I don’t think this would have to be true, but if dividend stocks were markedly better in this respect, I would guess that their advocates would have been bragging about it to the point of being annoying.
Finally, the model will assume that the difference between nominal and effective yield in Figure 1 accounts for all the losses an investor experiences when things go wrong. This might be overly optimistic, depending upon just how these events were accounted for in producing the figure. As I have emphasized before, minimizing losses surrounding dividend cuts is key to achieving good performance from any Dividend Growth Investing (DGI) portfolio.
Table Of Modeling Parameters
Putting all the above together produces Table 1, which has enough information to enable modeling of the various possibilities. Looking at this table and Figure 1, it makes sense to me to use 2.5%, 4.5%, and 6.5% as the forecast yields for the modeling, along with the non-payers, at 0%.
The table also shows a column for the Dividend Aristocrats and for large-capitalization eREITS, both discussed later.
Table 1. Parameters for modeling of dividend paying stocks. DAs stands for Dividend Aristocrats. The source of the DA data is here.
Results Of Portfolio Yield Variation
Now we consider the investor placing 500 k$ in DGI for 30 years, while withdrawing at some rate. Figure 2 shows a comparison of the final portfolio value, in 2019 dollars, for the three cases of dividend paying ordinary stocks having the same total return.
The differences in Figure 2 are not large by comparison with the statistical variations. The low-yield portfolio appears to perform better on average. The impact of the higher average price return apparently overcomes that of the lower dividend yield.
Note that all the 20th percentile limits in Figure 2 are similar. One has a very large chance of ending 30 years with more capital than one began with, even with a withdrawal rate of 8%. Beyond that, at a 10% withdrawal rate the median value of the portfolio remains above zero at 30 years.
Figure 2. Final portfolio value, in 2019 dollars, after an investor withdraws the percentage indicated each year for 30 years. The vertical bars show the range from 20th to 80th percentile of the distribution of results. The color corresponds to the indicated yield. Calculations by author.
Comparison With Stocks That Pay No Dividends
Figure 3 shows the comparison of the 4.5% portfolio with a portfolio of non-payers having the same total return. Here is where the Monte Carlo simulations pay off. The dividend payments do not fluctuate with the market. This limits the losses in down years. This ameliorates negative sequences of returns, and has the impact that portfolio depletions are much less common.
The upside for the portfolio of non-payers is quite high, but the downside is catastrophic. The 20th percentile limit gets stuck at zero as the withdrawal rate exceeds 6%. By 8%, median case for the non-payer is that the portfolio is depleted.
Figure 3. Final portfolio value, in 2019 dollars, after an investor withdraws the percentage indicated each year for 30 years. The vertical bars show the range from 20th to 80th percentile of the distribution of results. The blue curve corresponds to the 4.5% portfolio and the red curve to an S&P 500 index fund. Calculations by author.
In contrast, the dividend paying portfolio ends up above the initial invested capital in 80% of the cases at a withdrawal rate of 8%. Such a portfolio is unlikely to become exhausted even if 10% of the initial value of the funds is withdrawn per year.
At this point, we have seen that dividend-paying stocks strongly outperform non-payers in a retirement withdrawal scenario, even if they have the same total return. I don’t recall seeing evidence in support of this before. It may exist; I have not read widely in the DGI literature. The result is intuitively sensible. (But perhaps someone will find a mistake in the analysis.)
Comparison With The Dividend Aristocrats
In actuality, dividend payers have produced more total return than non-payers. We can look at the impact of this by comparing with the results of a model based on the Dividend Aristocrats, discussed in more detail here. These are companies that have increased their dividends each year for 25 years or more.
For the present model, we need an effective yield including the losses when stocks are dropped from the Dividend Aristocrat index. The Aristocrats boast a nominal 10.4% total return. If we incorporate the historical 6.6% dropout rate, and assume that 90% of the capital is recovered when a stock drops out, the effective total return becomes 9.7%.
Figure 4 shows a comparison of results from the 4.5% portfolio with those from a portfolio of the Dividend Aristocrats. The total return of the Aristocrats is only 1.3% larger than that of the higher-yield portfolio, but the price return is much larger, being 7.2% vs 4.1%. This dominates the results.
I find these results to be remarkable. For a 10% withdrawal rate, the higher yield of the Aristocrats pulls their 20th percentile result above the initial capital invested. The combination of reliably paying dividends and higher total return produces results that are far superior.
Figure 4. Final portfolio value, in 2019 dollars, after an investor withdraws the percentage indicated each year for 30 years. The vertical bars show the range from 20th to 80th percentile of the distribution of results. The blue again shows the 4.5% portfolio. The magenta shows results for a portfolio composed of the Dividend Aristocrats. Calculations by author.
Comparison With Large-Cap eREITs
Equity REITs, or eREITs, own real property from which they collect income. This is typically as rent. They are legally required to pay out 90% of their taxable income as dividends, and so overall pay large dividends compared to the average stock.
NAREIT maintains various indices of their performance over time. Here we use the All eREITs Index. The index is capitalization-weighted, and so reflects the behavior of the large-cap eREITs. We used results from 1998-2018 to find the parameters shown in Table 1.
Figure 5 shows a comparison of portfolios composed of large-cap eREITs and of Dividend Aristocrats. At low withdrawal rates, the eREITs produce median results comparable to the 2.5% portfolio of Figure 2, with better performance at the 20th percentile.
At higher withdrawal rates, the eREITs perform comparably to the Aristocrats, thanks to their higher yield. For withdrawal rates of 6% or more, they outperform all portfolios having lower total return.
Figure 5. Final portfolio value, in 2019 dollars, after an investor withdraws the percentage indicated each year for 30 years. The vertical bars show the range from 20th to 80th percentile of the distribution of results. The black shows results for a portfolio of large-capitalization equity REITs. The magenta shows results for a portfolio composed of the Dividend Aristocrats. Calculations by author.
Portfolio Implications And Opportunity
We’ve been in a fairly long phase of outperformance by growth stocks, many of which pay no dividends. Both the Dividend Aristocrats and large-cap eREITs have underperformed in the present decade.
Despite this, I would argue that they both have provided and do provide a safer direction for a retiree drawing steadily from a portfolio, and especially if the withdrawals are large. Other quality investments that pay dividends will have similar advantages.
An investor with a Dividend Growth mindset can continue to grow dividend yields by investing in these two areas and other quality securities. This can continue even while withdrawing funds to live on.
Over time, the performance of any such investment class will vary. This is one reason why diversification is essential. I recommend diversification across several asset classes, not just two.
It is particularly difficult to judge the future of large-cap eREITs at this moment. They have evolved during the past 20 years to have significantly lower leverage than they once did. Despite their larger leverage of the past, they have always had spectacularly low rates of bankruptcy. They appear to be very secure investments and perhaps they will remain priced accordingly.
One has good ways to invest in the broad markets of the Aristocrats and eREITs. There are several ETFs oriented to dividend aristocrats by some definition. Examples include NOBL, REGL, WDIV, and SDY. They have low expense ratios. There are also other ETFs oriented toward higher yielding stocks, such as Vanguard’s VHDYX.
In addition, there are ETFs oriented to the broad market of eREITs. Examples include the Vanguard REIT ETF (VNQ) and the iShares US Real Estate ETF (IYR). There are far worse retirement strategies than to place funds in a mix of these ETFs and to plan on withdrawing from them at a significant rate.
At High Yield Landlord, we believe that, even beyond the opportunities illustrated above, there is alpha to be found in the small-cap space for REITs and other real assets, and more secure income to be found in small-cap REIT preferred stocks. There are many firms too small to attract the interest of the large, institutional investors who drive the large-cap space. I look at the HYL portfolios and commentary for some of my investment ideas that seek to outperform even the best of the above examples.
With Better Information, You Get Better Results…
At High Yield Landlord, We spend 1000s of hours and well over $20,000 per year researching the REIT, MLP and BDC markets for the most profitable investment opportunities and share the results with you at a tiny fraction of the cost.
- •We are the #1 rated service on Seeking Alpha with a perfect 5/5 rating.
- •We are the #1 ranked service for Real Estate Investors with over 750 members.
Take advantage of the 2-week free trial and join our community of 750 "landlords" before we hike the price!
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.