The Markowitz Revolution

It wasn't until the middle of the century that Harry Markowitz, then a young, unknown professor at the University of Chicago, had the fundamental insight that it was correlation that mattered. In a then-little-noticed article published in a then-obscure journal,³ Markowitz not only demonstrated that correlation mattered (or “covariance” as he put it); he went far beyond that insight to show that by paying attention to correlation investors could create portfolios that reduced risk without necessarily reducing returns. Indeed, in some cases (such as adding international stocks to a U.S. stock portfolio) risk could be reduced while returns actually increased. And in every case in which portfolios were sensibly constructed, risk could be reduced faster than returns declined. Markowitz was the Einstein⁴ of the financial world, revolutionizing the way we thought about risk and return and making possible all the subsequent advances that have come to be known as modern portfolio theory.

This astonishing idea took more than 20 years to be adopted in the design of actual portfolios. Part of the delay had to do with the inherent time lag between ideas generated in the academy and the adoption of those ideas in the real world. But there was another problem; namely, the cost and unsophisticated nature of computing power. Consider how portfolios are designed using Markowitz's insights.

Asset allocation à la Markowitz seems deceptively simple. We need only know the future risks (expressed in terms of the standard deviation of returns) and returns for each asset class we wish to work with, as well as the future correlation of each asset class with each other asset class. This data is programmed into a computer which then chugs away, looking at all possible combinations of these assets.⁵ The computer program, which is an “optimizer,” is trying to identify those asset combinations that are optimal, that is, which produce the most return per unit of risk. Depending on how much risk we are willing to take, there may be many optimal portfolios available to us, and those will fall along the familiar efficient frontier curve. Possible portfolios that fall below the efficient frontier line are undesirable in the sense that they produce less return for the same risk. Portfolios that fall above the efficient market line are theoretically impossible.