Dividend Growth Sensitivity Study

I was asked about what happens to dividend growth projections when I limit my investigation to more recent times, starting in the mid-1950s.

There are differences. They are interesting.

Fortunately, they tend to reinforce, not undermine, our previous use of the data. I remain impressed by dividend-based strategies.

Regression Equations

Here are three sets of regression equations: the original set from 1921-1980 and new sets from 1956-1980 and 1956-1990. In these equations, y = the percentage growth of the dividend amount (not yield) and x = the dividend yield divided by the earnings yield using ten years of earnings, 100E10/P. This equals the payout ratio based on E10: that is, 100D/E10. As a rule, 100D/E10 is greater than the single-year payout ratio 100D/E because real earnings increase with time. Recently, the single-year payout ratio of the S&P500 has been between 30% and 40% while 100D/E10 has been between 40% and 50%.

From 1921-1980:

At year 4:
y = -0.3162x+23.799 plus 30% and minus 20%.
R-squared = 0.153.
When x = 40%, y = 11.2%.
When x = 60%, y = 4.8%.

At year 8:
y = -0.9914x+79.81 plus 60% and minus 30%.
R-squared = 0.2361.
When x = 40%, y = 40.2%.
When x = 60%, y = 20.3%.

At year 12:
y = -1.0328x+95.556 plus 100% and minus 50%.
R-squared = 0.1549.
When x = 40%, y = 54.2%.
When x = 60%, y = 33.6%.

From 1956 to 1980:

At year 4:
y = 0.0747x-3.3154 plus 9% and minus 8%.
R-squared = 0.0157.
When x = 40%, y = -0.3%.
When x = 60%, y = 1.2%.

At year 8:
y = -0.0303x+2.1357 plus 26% and minus 20%.
R-squared = 0.0003.
When x = 40%, y = 0.9%.
When x = 60%, y = 0.3%.

At year 12:
y = -0.199x+17.287 plus 40% and minus 20%.
R-squared = 0.009.
When x = 40%, y = 9.3%.
When x = 60%, y = 5.3%.

From 1956 to 1990:

At year 4:
y = -0.0039x+1.9347 plus 8% and minus 8%.
R-squared = 6E-05.
When x = 40%, y = 1.7%.
When x = 60%, y = 1.7%.

At year 8:
y = -0.2663x+20.927 plus 25% and minus 22%.
R-squared = 0.0328.
When x = 40%, y = 10.3%.
When x = 60%, y = 4.9%.

At year 12:
y = -0.6054x+43.278 plus 30% and minus 20%.
R-squared = 0.1055.
When x = 40%, y = 19.1%.
When x = 60%, y = 7.0%.

Typical Values

When x = 40%:

From 1921-1980:
Year 4: 11.2% (from -9% to 41%)
Year 8: 40.2% (from 10% to 100%)
Year 12: 54.2% (from 4% to 154%)

From 1956 to 1980:
Year 4: -0.3% (from -8% to 9%)
Year 8: 0.9% (from -19% to 27%)
Year 12: 9.3% (from -11% to 49%)

From 1956 to 1990:

Year 4: 1.7% (from -6% to 10%)
Year 8: 10.3% (from -12% to 35%)
Year 12: 19.1% (from -1% to 49%)

When x = 60%:

From 1921-1980:
Year 4: 4.8% (from -15% to 35%)
Year 8: 20.3% (from -10% to 80%)
Year 12: 33.6% (from -16% to 134%)

From 1956 to 1980:
Year 4: 1.2% (from -7% to 10%)
Year 8: 0.3% (from -20% to 26%)
Year 12: 5.3% (from -15% to 45%)

From 1956 to 1990:

Year 4: 1.7% (from -6% to 10%)
Year 8: 4.9% (from -17% to 30%)
Year 12: 7.0% (from -13% to 37%)

Previous Use of Dividend Growth Estimates

Previously, I used the 1921-1980 data. I argued that a 40% drop in (real) dividend amounts for the S&P500 index is unlikely going forward. My rationale was based on the current dividend payout ratio 100D/E10, which is around 40% to 50%. At today’s level, the worst case would be a permanent, (real) dividend cut of 20%.

Dividend-Based Design Example
What Do I Really Think About Dividends?

S&P500 dividend cuts of 40% have happened before. But single-year payout ratios have exceeded 100% as well (for the S&P500 index!).

Additional Observation

The data starting from 1956 is itself interesting because of its pessimism. During the high inflation that began in the 1960s, dividend growth fell behind inflation. Nominal dividends continued to grow. Real dividends fell.

This explains why the 1956-1980 series is the most pessimistic.

Conclusion

All of this gets back to earnings, real earnings averaged over an extended amount of time. When real earnings are high (relative to dividends), dividends are secure. They can grow. Currently, the S&P500 payout ratio is low. S&P500 dividends are secure.

Have fun.

John Walter Russell
May 6, 2006