Correlation

Correlation is expressed as the correlation coefficient (denoted by the Greek letter rho, ρ, or simply r), which ranges between -1 and +1. A correlation of +1 indicates that two assets move in perfect lockstep — when one rises 1%, the other always rises by a proportional amount. A correlation of -1 indicates perfect inverse movement — when one rises 1%, the other always falls by a proportional amount. A correlation of 0 indicates no linear relationship between the two assets' movements.

In Modern Portfolio Theory, as Harry Markowitz showed, the risk-reduction benefit of combining two assets in a portfolio is entirely determined by their correlation. When correlation is below +1, diversification reduces portfolio volatility. The lower the correlation, the greater the risk reduction. This is why the phrase 'don't put all your eggs in one basket' has a rigorous mathematical foundation: holding uncorrelated or negatively correlated assets can reduce total portfolio risk without sacrificing expected return.

Historically, U.S. stocks and investment-grade bonds have exhibited low or mildly negative correlations, which is the theoretical justification for balanced portfolios like the traditional 60/40 allocation. However, this relationship is not stable — during certain inflationary periods and market crises, stocks and bonds have moved together positively, reducing the diversification benefit. The 2022 market environment, when both stocks and bonds fell sharply as the Federal Reserve raised interest rates aggressively, was a notable example of correlation breakdown that caught many investors by surprise.

Correlation also has an important tendency to increase during market crises. In stress events, investors tend to sell all risky assets simultaneously regardless of their normal co-movement patterns — a phenomenon called 'correlation breakdown' or 'contagion.' This means that diversification benefits measured during normal markets often diminish precisely when they are needed most. Portfolio risk managers use stress testing and scenario analysis to assess portfolio behavior under high-correlation crisis environments.

The Pearson correlation coefficient — the standard measure — captures only linear relationships. Assets can have nonlinear dependencies (captured by tools like copulas) that the correlation coefficient misses. Additionally, correlation says nothing about causation or the magnitude of co-movements, only their direction and consistency. For practical portfolio construction, investors typically estimate correlations from rolling windows of historical returns and should be mindful that estimated correlations are inherently uncertain, particularly for assets with short or limited return histories.