Modern Portfolio Theory - Diversification and the Efficient Frontier

By EC Assets Research Team, Portfolio Research · Published · Updated

Modern Portfolio Theory: Modern Portfolio Theory, from Markowitz (1952), judges an asset by its contribution to whole-portfolio risk and return rather than in isolation. Because less-than-perfectly-correlated assets combine to lower volatility without sacrificing return, it formalises diversification and the efficient frontier - the foundation of quantitative asset allocation.

What Modern Portfolio Theory Says

Markowitz (1952) founded quantitative portfolio construction on one insight: an asset should be judged by what it contributes to the risk and return of the whole portfolio, not in isolation. A volatile asset that moves against the rest of the portfolio can reduce total risk. The investor's task is therefore to combine assets so that expected return is as high as possible for each level of risk.

The theory describes each portfolio with two numbers, expected return and variance. Expected return is the weighted average of the components; variance is not:

$$E[R_p] = \sum_i w_i,E[R_i], \qquad \sigma_p^2 = \sum_i \sum_j w_i w_j,\sigma_i \sigma_j,\rho_{ij}$$

When correlations are below one, portfolio volatility is less than the weighted average of the individual volatilities. That reduction, achieved without giving up expected return, is diversification.

How It Works

Plotting every attainable portfolio on a chart of volatility against expected return fills a region bounded on its upper-left edge by the efficient frontier, the portfolios with the highest return for each level of risk.

Adding a risk-free asset changes the picture. The best trade-off becomes a straight line from the risk-free rate that touches the frontier at the tangency portfolio, the risky mix with the highest Sharpe ratio. Tobin's separation result follows: every investor should hold the same risky portfolio and choose risk by mixing it with cash or borrowing. If all investors do so, the tangency portfolio is the market, which is the step to the CAPM.

Diversification has a limit set by correlation. With N assets of equal volatility σ and equal pairwise correlation ρ, an equally weighted portfolio has variance σ²(1/N + (1 - 1/N)ρ). At 20 percent volatility and a correlation of 0.3, ten assets give 12.17 percent, a hundred give 11.08 percent, and no number of assets gets below 10.95 percent: the correlated part of the risk cannot be diversified away.

In practice the optimiser's output depends on three estimated inputs: expected returns, volatilities and correlations. Expected returns are the hardest to estimate and the most influential, which is the root of the theory's practical problems. Even in the two-asset example below, raising A's expected return by one percentage point lifts its weight in the tangency portfolio from 59.5 to 66.7 percent; with many closely correlated assets the swings are far larger.

Worked Example

Asset A has an expected return of 8 percent and volatility of 12 percent; asset B 5 percent and 8 percent; their correlation is 0.2 and the risk-free rate 3 percent.

Weight in A Expected return Volatility Sharpe ratio
0% 5.00% 8.00% 0.25
25% 5.75% 7.22% 0.38
50% 6.50% 7.85% 0.45
75% 7.25% 9.60% 0.44
100% 8.00% 12.00% 0.42

Sharpe ratio = (expected return - 3%) / volatility.

The 50/50 mix has a volatility of √(0.5² × 12² + 0.5² × 8² + 2 × 0.5 × 0.5 × 12 × 8 × 0.2) = √61.6 ≈ 7.85 percent, against a weighted average of 10 percent. The minimum-variance mix holds about 26.4 percent in A and has 7.22 percent volatility, lower than either asset alone.

The tangency portfolio holds about 59.5 percent in A, with an expected return of 6.78 percent, volatility of 8.41 percent and a Sharpe ratio of 0.45. An investor who wants only 6 percent volatility holds about 71.4 percent in this portfolio and the rest in cash, for an expected return of 5.70 percent; no mix of A and B alone gets volatility that low. The correlation bench shows how the gains change with correlation.

When It Applies (and Limitations)

Inputs are uncertain. Michaud (1989) called mean-variance optimisers error maximisers: they load up on assets whose expected returns are overestimated and whose risks are underestimated. DeMiguel, Garlappi and Uppal (2009) found that across many data sets, optimised portfolios rarely beat simple equal weighting out of sample.

Variance is not the whole of risk. Variance treats gains and losses alike and ignores fat tails and skew. Investors who care about drawdowns supplement it with downside deviation and expected shortfall.

Correlations are unstable. They tend to rise in crises, so a frontier built on calm-period correlations overstates diversification in the moments that matter most.

Single period. The model chooses one portfolio for one horizon and assumes investors care only about mean and variance. Liabilities, rebalancing costs, taxes and changing opportunities over time all sit outside it.

Why It Matters for Institutional Investors

The shared vocabulary. Efficient frontiers, diversification, the tangency portfolio and the risk-return trade-off all come from the theory, and strategic asset allocation for pensions, endowments and sovereign funds still starts from it.

Robust implementation. Black and Litterman (1992) blend market-implied returns with investor views to tame input sensitivity; shrinkage, weight limits and resampling serve the same purpose. Risk-based approaches such as risk parity avoid expected-return estimates altogether.

Judging allocations. Asking what an asset adds to the existing portfolio, rather than how attractive it looks alone, is the most practical lesson of the theory and the basis of sound portfolio construction.

Knowing the limits. An allocator's sophistication shows in how they handle estimation error, non-normal risk and unstable correlations, not in whether they can draw the frontier.

References

  1. Markowitz, H. (1952). Portfolio Selection. Journal of Finance, 7(1), 77-91.
  2. Tobin, J. (1958). Liquidity Preference as Behavior Towards Risk. Review of Economic Studies, 25(2), 65-86.
  3. Michaud, R. O. (1989). The Markowitz Optimization Enigma: Is 'Optimized' Optimal? Financial Analysts Journal, 45(1), 31-42.
  4. Black, F., & Litterman, R. (1992). Global Portfolio Optimization. Financial Analysts Journal, 48(5), 28-43.
  5. DeMiguel, V., Garlappi, L., & Uppal, R. (2009). Optimal Versus Naive Diversification: How Inefficient Is the 1/N Portfolio Strategy? Review of Financial Studies, 22(5), 1915-1953.

Frequently asked questions

What is the core idea of Modern Portfolio Theory?

That an asset should be evaluated by what it adds to the portfolio's overall risk and return rather than on its own. Because assets that do not move together partly offset each other, combining them can lower total volatility without reducing expected return.

What is the efficient frontier?

The set of portfolios that offer the highest expected return for each level of risk, or the lowest risk for each level of return. Any portfolio below it is inefficient, because another portfolio earns more for the same risk or takes less risk for the same return.

Why is diversification called a free lunch?

Because combining imperfectly correlated assets reduces portfolio volatility without lowering the weighted expected return. It is the one improvement in portfolio construction that does not require giving something up, as long as the correlation assumptions hold.

What are the main weaknesses of Modern Portfolio Theory?

Optimised portfolios are extremely sensitive to estimated inputs and tend to concentrate in assets with overestimated returns; variance ignores fat tails and skew; and correlations rise in crises, when diversification is most needed. DeMiguel, Garlappi and Uppal (2009) find that naive equal weighting often beats optimised portfolios out of sample.

Is Modern Portfolio Theory still used?

Yes, as the vocabulary and starting point of asset allocation. Practitioners rarely feed raw estimates into an optimiser, though: they add constraints, shrink estimates, blend in views with Black-Litterman, use risk-based methods such as risk parity and supplement variance with downside and tail measures.

Stay informed

Market commentary, firm news and research from EC Assets - direct to your inbox.