Definition
Correlation is a measure of how consistently two sets of values move together, ranging from a coefficient of one for values that move perfectly in step to a coefficient of negative one for values that move perfectly opposite, with zero meaning no consistent relationship.
Why it matters
Correlation is the basis that underlies diversification. Combining assets whose returns are not perfectly correlated produces a portfolio whose variability is lower than the average of its components' individual variabilities. Without a way to summarize how consistently pairs of variables move together, there is no analytical basis for portfolio construction or for measuring how much diversification a given combination actually delivers.
How it works
Correlation between two sets of values is computed by measuring how far each pair of matched values falls from their respective averages, taking the product of those two deviations for each pair, averaging the products, and dividing by the product of the two standard deviations. The result is a coefficient bounded between minus one and plus one. A coefficient of one means the two variables always move together in fixed proportion; minus one means they always move in opposite directions in fixed proportion; zero means there is no consistent linear relationship between them. Correlation captures only the linear component of a relationship, so two variables with a strong but non-linear relationship (for example a curved or threshold-based dependence) can show a correlation coefficient near zero even when they are not independent. Worked example: if U.S. large-cap equity returns and U.S. investment-grade bond returns show a correlation coefficient of around zero point two over a given historical period, this indicates a mild positive tendency to move together, considerably weaker than the near-one correlation typically observed between large-cap equity and small-cap equity returns over the same period.
In practice
For an individual constructing a retirement portfolio, correlation is the analytical property that determines whether diversifying across asset classes actually reduces variability. If two funds have a correlation near one, holding both provides little diversification benefit; if two funds have a correlation near zero or negative, holding both produces meaningful variability reduction at any given expected return. A professional working with portfolio construction examines historical correlations, though correlations shift over time and tend to rise during periods of market stress, a limitation of correlation-based diversification that is particularly relevant for participants near or in retirement. Plan fiduciaries reviewing target date fund construction and default investment allocations should recognize that stated diversification benefits are conditional on the correlation assumptions used to build the glide path.
Related terms
- Standard deviation
- Variance
- Volatility drag
- Risk-adjusted return
- Sharpe ratio
- Monte Carlo simulation
- Geometric Brownian motion
- Sequence of returns risk