So, diversifying into various investments is clearly a smart thing, and in general leads to lower volatility for our portfolios. But, there are two common misunderstandings when it comes to diversification:
Correlations between investment returns are not static and in reality change frequently. The principle of diversification is based on the concept of correlations between investment returns. That is if, based on historical data, IBM stock moves up in price when Wal-Mart stock moves down in price, and I believe this relationship will hold in the future, buying both IBM and Wal-Mart stock will provide diversification to my portfolio because their prices don’t move together (hence the value of my portfolio is less volatile). Now, I don’t want them to move exaclty the opposite in price so that I don’t earn any positive return over time. What I want is for both of them to have a positive expected return and still move in different ways to offset each other so that over time I have a positive return but less volatility. The problem is, the correlation between IBM and Wal-Mart that I established using historical data may not hold in the future, and in fact is unlikely to hold. So, I think I’m getting diversification but I’m not, and when a major market event happens that crushes the price of IBM stock, it may also crush the price of Wal-Mart stock and I’m left exactly where I would have been without Wal-Mart. What’s important to know is that correlations are much more persistent on an asset class level, and to a lesser extent on a sector level, than on an individual stock or bond level. That is, the correlation between the stock and bond markets or the correlation between the general Information Technology (“IT”)