How to calculate the geometric mean and CAGR lets you find the appropriate average when data represent growth rates, percentage returns, multiplicative factors, or quantities that compound over time. The formula is GM = (x₁ × x₂ × ... × xₙ)^(1/n): the n-th root of the product of n positive values. Equivalent: GM = exp(mean(ln(xᵢ))). Always holds HM ≤ GM ≤ AM (HM-GM-AM inequality, Cauchy 1821), with equality only for identical data. The CAGR (Compound Annual Growth Rate) is a financial application of GM on annual returns, used by SEC, Federal Reserve, and per NIST statistical handbooks.

Formula and definition

Given n positive observations x₁, x₂, ..., xₙ, the geometric mean is: GM = (x₁ × x₂ × ... × xₙ)^(1/n). For large data calculate with logarithms (Lederman & Wisdom 1971): GM = exp((Σ ln(xᵢ)) / n), avoiding numerical overflow. For two values a, b: GM = √(a × b) (classical geometric mean, Pythagoras).

CAGR derives from GM: CAGR = (V_final / V_initial)^(1/n_years) - 1, where n_years is number of periods (not values). Example: investment $10,000 grown to $15,000 in 5 years: CAGR = (15,000/10,000)^(1/5) - 1 = 1.5^0.2 - 1 = 0.0845 = 8.45% annual. Not to be confused with arithmetic mean of annual returns (systematically overestimates). Used by SEC for fund prospectuses, Federal Reserve for GDP growth, NASA for population modeling, NIST for measurement statistics.

History and applications

The geometric mean was studied by Pythagoreans (6th century BCE) in musical proportions. Archytas of Tarentum called it "second mean". Cauchy (1821) and Schwarz rigorously proved the HM ≤ GM ≤ AM inequality. Hardy-Littlewood-Pólya (1934) generalized it in the power mean concept.

Modern applications: SEC requires CAGR in mutual fund prospectuses; Federal Reserve calculates GDP growth using GM; CDC publishes geometric population growth; NHS UK uses GM in epidemiological modeling. NIST uses GM in measurement statistics. NASA uses GM in space-time CAGR for satellite life cycles. CDC and EPA apply GM for water quality indicators (geometric mean of fecal coliform counts per ASTM D5392). The IRS uses geometric mean for inflation adjustments per IRC §1(f). Comparable Italian use: Banca d'Italia, ISTAT, COVIP.

Comparison table arithmetic/geometric/harmonic means

For close values (e.g., growth rates 5-12%) GM ≈ AM ≈ HM with divergence <0.1%. For dispersed values (e.g., 1 and 100) the three means diverge drastically: GM is the "true" multiplicative average per NIST guidance.

5 verified numerical examples

1. 5-year investment CAGR: initial capital $10,000, final $15,000. CAGR = (1.5)^0.2 - 1 = 8.45% annual. Arithmetic mean of annual returns (10.15%) overestimates by 20%.
2. Pension fund annual returns: 5 years with +5%, +8%, -3%, +12%, +6%. Factors 1.05, 1.08, 0.97, 1.12, 1.06. GM = (1.05×1.08×0.97×1.12×1.06)^0.2 = 1.0552. Compound average return = 5.52%/year. Per SEC fund prospectus standards.
3. US GDP growth 5 years: annual rates 2.3%, 2.9%, -3.4% (2020 COVID), 5.7% (2021), 2.5%. GM = ((1.023×1.029×0.966×1.057×1.025)^0.2 - 1) = 2.0%/year (Federal Reserve, BEA data).
4. Geometric mean of ratios: P/E ratios 15, 25, 50 of 3 stocks. GM = (15×25×50)^(1/3) = (18,750)^(1/3) = 26.57. More representative than arithmetic (30) for aggregated ratios per NYSE/NASDAQ index calculations.
5. Cell population growth: 100 → 250 → 500 → 1,200 in 3 hours. Hourly CAGR = (1,200/100)^(1/3) - 1 = 1.289 - 1 = 28.9% per hour. Typical in microbiological models per CDC/FDA guidelines.

Model limitations

• Only positive values: GM requires x_i > 0 to be defined (root of product). For data with negative values (financial losses ≤ -100%) calculation is undefined. Solution: transform rates to factors (1 + r) and apply GM to factors, then subtract 1, per NIST measurement standards.
• Small outlier sensitivity: a value near zero dominates GM. Example: GM(1, 100) = 10 vs AM 50.5. Clean data or use trimmed GM (excluding top/bottom 5%).
• Contextual interpretation: GM is appropriate for multiplicative data (rates, ratios, factors); not for additive sums ($, kg, people).

Common errors

• Using AM for returns: for compound financial returns, AM systematically overestimates. CAGR (geometric) is correct measure per SEC standards.
• Confusing GM with product: GM is n-th root of product, not product itself.
• Including zeros or negatives: GM is undefined; filter or transform.
• CAGR on 1 year: formula requires n ≥ 2 periods. For 1 year growth = (V_final/V_initial - 1).
• Not distinguishing factors and rates: factor is (1 + rate). GM of factors - 1 = compound average return.
• Applying to additive quantities: for summing $, kg, populations use AM, not GM per NIST guidelines.

Numbered practical applications

1. Stock investment CAGR calculation: portfolio $50,000 grown to $75,000 in 7 years. CAGR = (75/50)^(1/7) - 1 = 6.0%/year. Standard for fund performance reporting per SEC Rule 482.
2. 5-year company revenue growth: $100,000 → $180,000. CAGR = 12.5%/year. Used in business plans and corporate valuations per Federal Reserve and SEC.
3. Population growth rate: US Census Bureau calculates demographic CAGR by state/county. US 2010-2020: +0.7%/year average growth.
4. Pension fund comparison: SEC/IRS publish CAGR 5/10 years of mutual funds. Vanguard target-date 10y CAGR ~7-8%, S&P 500 historical CAGR ~10% (1928-2024 BLS data).
5. Microbiological growth: microbiology labs (e.g., CDC) calculate CAGR of bacterial populations during incubation, virulence indicator per ASTM D5392.
6. Price indices: BLS calculates CPI (Consumer Price Index) using geometric mean of elementary aggregates weighted by spending shares per BLS Handbook of Methods.
7. Investment strategy comparison: backtesting with Sharpe-adjusted CAGR. Superior to arithmetic mean for personal financial planning per FINRA standards.

For technical support: info@calcoi.com.

Updated July 2026