Please Note: Blog posts are not selected, edited or screened by Seeking Alpha editors.

A Model For Dividend Growth Stock Buy Decision-Making


This blog post describes a model based on one described by a fellow Seeking Alpha commenter.

It adheres to the spirit of the described method but changes it in a couple of key respects.

I present the relative attractiveness of buying the stocks in my portfolio.

Recently I asked the community of dividend growth investors in Ron Smith's DGR Chit Chat blog how they go about making the decision of when to buy shares of the stocks they know they want to add.  Seeking Alpha member "goodies" responded with a method employing these inputs:

I thought this was an interesting approach, so I decided to play around with it, see how it worked for my portfolio, and then tailored it to my own needs.  Below is a description of the model I came up with.

Dividend Growth Stock Buy Decision Model

For this model, I decided to include dividend PAAY (percent above average yield), 5-year dividend compound annual growth rate, fair value, and projected growth.  I describe each below so that interested readers can offer critiques and/or tailor it to their needs and preferences.

  • Dividend PAAY.  This input comes from M* and compares the yield available today with the company's average yield over the last 5 years.  Here's a screenshot of the relevant information for Microsoft:

I calculated the percent that each stock's current yield deviates from its 5-year average, determined the average and standard deviations for the portfolio, and then awarded points:  2 points for +1 standard deviation (that would be yields that are somewhat higher now than the 5-year average), 3 points for +2 standard deviations (much higher than the 5-year average), and zero points for negative figures and outliers -- companies whose current yield is lower than the 5-year average or whose current yield is wildly higher than the 5-year average.

  • 5-Year Dividend CAGR.  Because I'm interested in dividend growth, I decided to add the company's recent dividend growth history to the model.  In the vast majority of cases, I used the CCC list's figures; in the few cases of stocks in my portfolio not on the CCC list, I used FAST Graphs figures.  Again I awarded points based on standard deviations: zero for outliers, 1 for -1 standard deviation, 2 for +1 standard deviation, and 3 for +2 standard deviations. The average 5-year DGR in my portfolio was about 10.7%; stocks with a 5-yr DGR as low as 4% were awarded 1 point.  Stocks that have had very high dividend growth over the last 5 years, like SBUX, LOW, and HD, were awarded 3 points.

  • Fair Value.  I calculated an average fair value using 3 inputs:  M*'s fair value, the fair value represented by FAST Graph forecasting tab using the "normal multiple" option (which is a 5-year chart), and the fair value imputed to the 5-year average yield.  I calculate the latter as FV = current price/(current yield/5-year average yield).  I averaged these 3 methods for determining fair value, then calculated the percent it deviated from current price.  I awarded zero points for any stock undervalued by more than 25% (which often indicates turnaround and speculative plays); 1 point for stocks undervalued by 15-25%; 2 points for stocks at value or undervalued by up to 15%; zero points for stocks overvalued by up to 10%; and I decided to award -4 points for any stock overvalued by more than 10%.  This was a way to make sure that well overvalued stocks don't make their way into the top tier for buying.

  • Projected growth.  I used FAST Graphs forecasting to determine analysts' projected growth for the company, on the theory that future dividend raises will come from the company's future growth. In the Forecasting section, it's listed in a green box labeled "Estimated Adjusted Earnings Growth"; the following screenshot is for MSFT.

I awarded 2 points for growth >10%, 1 point for 4-10% growth, and zero points for anything lower.  I didn't award negative points for negative projected growth.

For my own purposes, I also add in a factor to account for how much below my target allocation I'm at with each holding, but to generalize it to other interested investors, I took that out.  I collected the relevant information on July 13. The maximum point total of this model is 10, and the minimum is -4.  Theoretically, if you agree with the rationale for awarding points, then the higher the score, the potentially more interesting a stock is for purchase now.  Below is a table of results (points awarded only, not all the inputs; the table with inputs is pretty big).  Comments and critiques are welcomed.