Estimating Growth From Dividend History

Why Measure Growth From Dividend History?

A company’s past dividends provide observable evidence about how its distributions have changed. That evidence can be summarized in more than one way: the arithmetic average of annual percentage changes describes the average rate across observed years, while compound annual growth describes the single annual rate that would connect the first and last payments. These measures answer different questions and need not agree.

Neither calculation, by itself, establishes what the company will pay in the future. A historical growth rate is an input to judgment, not a guarantee. For a forecast to be defensible, the analyst should consider the length and consistency of the record, unusual payments, and whether the company’s circumstances and dividend policy are likely to persist.

Definition: For positive annual dividends $D_0,D_1,\ldots,D_n$, the growth rate from year $t-1$ to year $t$ is $g_t=(D_t-D_{t-1})/D_{t-1}=D_t/D_{t-1}-1$. The arithmetic average of the $n$ annual rates is $\bar{g}_{A}=(g_1+\cdots+g_n)/n$. The compound annual growth rate (CAGR) over the $n$-year interval is $g_C=(D_n/D_0)^{1/n}-1$.

The denominator in an annual growth rate is the dividend in the preceding year. Thus, a change of $0.10 is 10% growth when the prior dividend is $1.00, but only 5% growth when the prior dividend is $2.00. A percentage rate describes a relative change, not just a dollar increase.

Arithmetic Average and Compound Growth

The arithmetic average gives equal weight to each observed annual percentage rate. It answers: “What was the average of the annual rates in this record?” It does not generally give the rate that connects the starting dividend to the ending dividend.

CAGR answers a different question: “What constant annual rate would produce the observed ending dividend if applied to the starting dividend for the full period?” It is an endpoint summary. It does not claim that the dividend actually grew at that same rate in every year.

D_n=D_0\prod_{t=1}^{n}(1+g_t)

This identity explains why arithmetic average growth and CAGR can differ. The annual rates compound by multiplication, not by addition. If one annual rate is negative, it reduces the dividend base on which later growth applies. Averaging the rates alone does not preserve that sequence of changes.

Why the CAGR Formula Holds

Suppose the starting dividend is positive and the observed record spans $n$ annual intervals. If one constant rate $g_C$ were applied each year, the dividend after one year would be $D_0(1+g_C)$, after two years $D_0(1+g_C)^2$, and after $n$ years $D_0(1+g_C)^n$. To match the observed ending dividend, set that expression equal to $D_n$ and solve:

D_0(1+g_C)^n=D_n \quad\Longrightarrow\quad (1+g_C)^n=\frac{D_n}{D_0} \quad\Longrightarrow\quad g_C=\left(\frac{D_n}{D_0}\right)^{1/n}-1.

The exponent is the number of annual intervals, not the number of dividend observations. Four annual payments usually span three intervals. Using four as the exponent in that case understates the calculated annualized rate.

Worked Example: Comparing the Two Growth Summaries

A hypothetical firm paid annual dividends of $1.00, $1.20, $1.08, and $1.296 over three consecutive annual intervals. Calculate the arithmetic average of the annual growth rates and the CAGR. State the method and identify a common mistake.

Method: calculate interval rates first, then calculate the endpoint rate separately. The three annual rates are:

g_1=\frac{1.20}{1.00}-1=0.20=20\%, \qquad g_2=\frac{1.08}{1.20}-1=-0.10=-10\%, \qquad g_3=\frac{1.296}{1.08}-1=0.20=20\%.

The arithmetic average is $(20\%-10\%+20\%)/3=10\%$. The record has three intervals, and its endpoint ratio is $1.296/1.00=1.296$. Therefore:

g_C=(1.296)^{1/3}-1\approx 0.0903=9.03\%.

The check is that compounding the actual annual factors gives $1.20\times0.90\times1.20=1.296$, exactly the endpoint ratio. The constant CAGR is slightly below 10% because it must reproduce the endpoint after compounding. The common mistake is to report the arithmetic average, 10%, as the rate that links the endpoints; applying 10% for all three years would give $1.00(1.10)^3=1.331$, not $1.296.

Working With Dividend Histories

A careful calculation starts by organizing the observations in chronological order and confirming that each amount represents the same type of payment. Annual regular dividends should not be mixed casually with special dividends, stock splits, or payments from a different reporting period. Then count the intervals, compute each annual rate when the arithmetic average is needed, and calculate CAGR from the endpoints when a compound summary is needed.

Check the observations.
Confirm the dividend amounts, the dates they cover, and whether they are regular per-share payments stated on a comparable basis.

Count the intervals.
For $n+1$ consecutive annual observations, there are $n$ annual growth intervals.

Choose the measure.
Use the arithmetic average to summarize observed annual rates; use CAGR to summarize the annualized change between endpoints.

Evaluate the pattern.
Inspect annual changes and the business context before deciding whether either summary is a reasonable forecast guide.

If annual observations are missing, the endpoint formula can still annualize the change when the elapsed time is known. For example, if payments are recorded only at the start and end of a four-year period, use four intervals in the exponent. Do not treat the two observations as one annual interval. If the exact elapsed period is unknown, an annualized rate cannot be calculated reliably from endpoints alone.

Worked Example: A Dividend Cut Within an Otherwise Positive Record

A hypothetical insurer paid annual dividends of $0.50, $0.55, $0.605, $0.484, and $0.5324 in five consecutive years. Calculate the arithmetic average annual growth and the four-year CAGR. Explain what the difference says about the record, without assuming the history will repeat.

Method: calculate the four year-to-year rates and then the endpoint CAGR. The annual rates are:

\frac{0.55}{0.50}-1=10\%,\quad \frac{0.605}{0.55}-1=10\%,\quad \frac{0.484}{0.605}-1=-20\%,\quad \frac{0.5324}{0.484}-1=10\%.

The arithmetic average is $(10\%+10\%-20\%+10\%)/4=2.50\%$. There are four intervals between the first and fifth observations. The endpoint ratio is $0.5324/0.50=1.0648$, so:

g_C=(1.0648)^{1/4}-1\approx 0.0158=1.58\%.

The result checks against the total change: the last payment is 6.48% above the first, and annualizing that increase over four years gives about 1.58% per year. Both summaries reflect the dividend cut, but they do so differently. The common mistake is to omit the negative year from the arithmetic average because the other three rates are positive. That would give 10%, which does not describe the full record. Before using either rate as a forecast, an analyst would investigate whether the cut was temporary, reflected a lasting change in earnings or policy, or coincided with another relevant event.

When Is Historical Growth a Defensible Forecast?

A calculated historical rate becomes a forecast only after an argument is made that the conditions behind the past pattern remain relevant. A longer record can reduce the influence of one unusual year, but it can also blend together distinct business periods or dividend policies. A short, smooth record may be easy to summarize but may provide little evidence that the pattern will continue.

Useful questions include:

Is the payment series comparable? Check for special dividends, changes in the number of shares, splits, and differences in whether the amounts are annual or quarterly.
Is the pattern stable? Compare annual rates and, where appropriate, subperiod growth rates. A large difference may indicate a change in trend rather than random variation around one rate.
Were there unusual events? A dividend cut, rebound, or one-time distribution can strongly affect an arithmetic average or endpoint CAGR.
Does the business support the pattern? Consider whether earnings, cash generation, and dividend policy could plausibly sustain the historical pace. Past distributions alone do not establish future capacity.
Is the proposed rate meant to describe a near-term period or a longer horizon? A recent subperiod may be more relevant to a near-term forecast, while a longer period may better show how the company behaves across different conditions. Neither choice is automatic.

The earlier material on dividend growth from retention and return on equity provides a different way to reason about growth: it relates growth to how much earnings are retained and the return earned on reinvested funds. Historical dividend growth is an observed outcome. A difference between the historical rate and a rate supported by business fundamentals is a reason to investigate, not a reason to force the estimates to match.

Worked Example: Detecting a Change in the Recent Growth Pattern

A hypothetical utility paid annual dividends of $0.80, $0.88, $0.968, $1.0164, and $1.06722 over four intervals. Calculate the arithmetic average annual growth, the full-period CAGR, and the CAGR over the most recent two intervals. Then assess which historical summary is more informative for a forecast if the two latest increases reflect a newly adopted, still-uncertain policy.

Method: compare annual rates with endpoint CAGRs over the full and recent periods. The annual rates are $10\%$, $10\%$, $5\%$, and $5\%$. Thus, the arithmetic average is $(10\%+10\%+5\%+5\%)/4=7.50\%$.

For the full four-interval period, the endpoint ratio is $1.06722/0.80=1.334025$. Since $1.155^2=1.334025$, the fourth root is $\sqrt{1.155}$, approximately $1.0747$. Therefore the full-period CAGR is about $7.47\%$.

For the most recent two intervals, the ratio is $1.06722/0.968=1.1025$. Since $\sqrt{1.1025}=1.05$, the recent-period CAGR is exactly $5.00\%$. These calculations check against the stated annual rates: the latest two dividends each grew by 5%, while the full history includes two earlier increases of 10%.

If the policy change is real but uncertain, the recent 5% rate may be more relevant to the current pattern, while the full-period 7.47% rate documents the longer record. Neither should be selected mechanically as a forecast. The common mistake is to call 7.50% the current trend merely because it is the arithmetic average; it blends two distinct observed phases and does not resolve whether the new policy will persist.

Limits and Common Traps

Percentage growth requires a nonzero prior dividend. If the prior dividend is zero, the usual percentage growth formula divides by zero and is undefined. If the starting dividend is zero, the ordinary endpoint CAGR is also undefined. If dividends are negative, the standard positive-growth interpretation of CAGR does not apply. In such cases, report dollar changes or use a clearly specified alternative analysis rather than presenting an ordinary CAGR.

Rounding can also obscure a result. Keep intermediate rates unrounded when calculating the arithmetic average, and retain sufficient precision when taking a root for CAGR. Round only the final reported rate, with the precision appropriate to the context.

Situation	Appropriate interpretation	Common trap
Several annual rates are observed	Their arithmetic average summarizes those rates	Assuming it must equal the endpoint CAGR
Only endpoints over a known span are used	CAGR annualizes the endpoint change over that span	Using the number of observations instead of intervals
A zero or negative dividend occurs	Check whether percentage growth is defined before calculating	Reporting a standard growth rate despite a zero denominator
A cut or special payment appears	Investigate its cause and comparability with regular payments	Excluding an inconvenient observation without justification

Key takeaway: The arithmetic average summarizes the observed annual percentage changes; CAGR is the constant annual rate connecting the endpoints. Both are historical descriptions. A forecast requires a separate judgment about comparability, stability, and the likelihood that the underlying conditions will continue.

Check Your Understanding

For each problem, identify the relevant number of annual intervals and distinguish an arithmetic average from an endpoint CAGR where requested.

A hypothetical company’s annual dividends are $2.00, $2.10, $2.31, and $2.541. Calculate the arithmetic average of its three annual growth rates. (Answer: 8.3333%)
A hypothetical manufacturer’s annual dividends are $0.75, $0.90, $0.81, and $0.972. Calculate the arithmetic average annual growth and the three-interval CAGR. (Answer: arithmetic average 10.00%; CAGR approximately 9.03%)
A hypothetical fund paid $1.20 per share four years ago and $1.50 per share now. Calculate the endpoint CAGR, assuming the endpoints are four years apart. (Answer: approximately 5.7371%)
A hypothetical company’s annual dividends are $2.00, $2.20, $2.20, and $2.42. Calculate the arithmetic average annual growth and the three-interval CAGR. (Answer: arithmetic average 6.6667%; CAGR approximately 6.56%)
A hypothetical new issuer paid no dividend in its first observed year and $0.12 in the next year. Can the ordinary percentage growth rate from the first year to the second, or a standard endpoint CAGR starting from the first year, be calculated? (Answer: No; each formula has a zero starting dividend in its denominator or endpoint ratio, so the percentage rate is undefined.)

What You'll Learn