Despite what some investors believe, it is not the individual stocks or bonds in their portfolios that drive returns. Anne Kennedy explains that instead, the most important influence on long-term portfolio returns is the mix of assets.

Studies have shown that asset mix accounts for over 85% of long-term returns. For anyone managing a portfolio, for themselves or as part of their job, that makes asset allocation a concept worth understanding.

Asset allocation simply refers to a portfolio's mix across broad asset classes: stocks, bonds, cash and other investments, such as real estate or commodities. It also refers to the mix of investments within each asset class. In the stock portion of a portfolio, asset allocation refers to the blend of foreign, domestic, large company and small company stocks, as well as stocks in various equity disciplines such as growth or value.

The goal of asset allocation is to help investors maximise their return forany given level of risk. It accomplishes this goal by determining the appropriate diversification across and within asset classes. Appropriate diversification comes not just from owning different assets but also from owning assets that respond differently to the economic and investment environment. Therefore, effective asset allocation tools have to take into account how various assets behave over time, both individually andtogether. Do the returns on stocks and bonds tend to rise and fall together or are stock returns rising while bond returns are falling? Knowing whether the returns on assets move together (positive correlation) or in an opposite pattern (negative correlation) is important in constructing the right mix of assets.The importance of asset allocation and diversification has been understoodfor a long time, at least intuitively. Until theory and the power of computers made it possible to quantify that intuition, there was no effective tool to help investors determine the appropriate mix of assets. Harry Markowitz was the first one to seize the power of computers and provide investors with a quantitative tool for asset allocation decisions.

This approach made the following assumptions:
• Assets are not perfectly correlated;
• Investors are risk averse;
• Investment returns are normally distributed; and
• Risk is defined as standard deviation of returns.

His model produced the expected returns of assets (based on historical data) as well as the relationship or correlation among those assets' returns. He then used these results to plot all of the asset mixes that would maximise returns for any given level of risk. The plot or curve that resulted was called the “efficient frontier”. Markowitz's effort, known as modern portfolio theory, won him a Nobel Prize in economics. Modern portfolio theory continues to be the standard approach to asset allocation. As more work has been done on asset allocation, however, some weaknesses in this approach have been identified.

The asset allocation outcomes are not driven by client objectives. Themodel's quantifiable inputs define the efficient frontier - the recommended or optimal asset mixes among which to choose. These inputs include historical returns for various assets and measures of the volatility of these assets. Investors can only choose among the points onthe efficient frontier.

The process defines risk as any deviation from the expected return. Both upside and downside risk are penalised equally. Most investors, however, tend to penalise performance that fails to meet their goal more than they reward performance that exceeds their goal.

Improvements
Markowitz's method assumed that returns follow a symmetrical bell-shapedpattern - in other words, that they are normally distributed. This makes the math easy, but it does not describe the behaviour of investment returns in the real world. There has been a great deal of effort in recent years to build upon and improve the pioneering work done by Harry Markowitz. Frank Sortino has been instrumental in helping to develop what is known as post-modern portfoliotheory (PMPT), which is based upon the following assumptions:
• Assets are not perfectly correlated;
• Investors are risk averse;
• Investment returns are not normally distributed; and
• Risk is defined as downside deviation (described below).

Three major differences are obvious: PMPT does not assume that investmentreturns are normally distributed. This means that PMPT will reward investments that are skewed toward positive surprises for investors (upside deviations) and penalise investments that are skewed toward negative surprises or downside deviations from their target. This one difference initself makes asset allocation decisions more relevant to investors.

In PMPT, risk is not defined as both upside and downside deviations. Riskis identified as returns that fall below the targeted level. This can result in a significantly different asset mix - one that can help investors maximise risk-adjusted returns.

With PMPT, an investor's specific investment goals are part of the process of defining the efficient frontier. An investor who can take on risk and needs a fairly high rate of return will have a very different efficient frontier than an investor who wants little or no risk and is primarily concerned with preserving capital. With modern portfolio theory, both investors would choose a different point on the same efficient frontier.

Customising an efficient frontier to meet the specific needs of an investor is an important part of an effective investment strategy. For example, cash is usually considered a “risk-free” investment. If a client's investment goal is to earn an 8% annual return, cash becomes a very risky investment. In fact, at current interest rates, there is close to a zero percent probability that an all-cash portfolio will meet the investor's goal.

Stocks are more volatile and, therefore, riskier in the short term.

However, over the longer term, they are actually a safer investment than cash for an investor whose objective is an 8% annual return. The point is that the true risk of any asset allocation strategy cannot be defined without knowing an investor's goals and objectives.

Implementation
How can the post-modern portfolio theory asset allocation be implemented?
There are five key factors that form the basis of an asset allocationrecommendation. Four are quantifiable factors, including:
• The expected return for each asset;
• The shape of the actual distribution of returns for each asset;
• Risk (defined as downside deviations from the client's target return);and
• The relationships (correlations) among the assets.

Once this quantitative exercise is completed, the results are combined with specific investor inputs. An investor should be able to answer the following questions:
• What is the investment time horizon for the pool of assets? How longwill the assets be invested?
• What are the specific investment objectives for this pool of assets?Consider capital preservation, liquidity needs, capital appreciation andannual income goals.
• What is the investor's risk tolerance? What amount of fluctuation ofquarterly and annual returns can the investor accept?
• Are there any special constraints? For example, is there a maximumdownward fluctuation in returns that the investment portfolio can tolerate?Now the quantitative inputs can be combined with the investor's specificneeds and objectives. The result should be the combination of assets that maximises the probability of meeting the investor's objectives with thelowest level of risk.

This process, of course, needs to be reviewed on a regular basis. Inputs will need to be changed when the investor's objectives change or when there has been some external shock or long-term shift in the fundamental investment inputs. The key point here is long-term. Asset allocation is designed for investors with a longer-term time horizon. Changing asset allocation strategy with every up and down in the financial markets defeats the purpose of the asset allocation strategy.

One final step is for investors to ensure that any investment manager they have hired adheres to the discipline for which they were selected. If an investment manger strays from their discipline, whether they are underperforming or outperforming, they can seriously distort the asset allocation process. It is very important, therefore, that asset managers understand their role in the overall asset mix and know that they will be judged not only on performance but also on their adherence to their discipline.

An important part of reviewing investment managers is to make sure that they are being measured against an appropriate benchmark. Measuring a small capitalisation stock manager against the large capitalisation S&P 500 Index is clearly not appropriate. A more relevant benchmark would be the Russell 2000 Index, which is based on a universe of small company stocks.

Asset allocation is the primary driver of a portfolio's return. It is, therefore, the most important part of portfolio construction. A good asset allocation policy will do more than identify asset combinations that maximise returns for a given level of risk. It will identify those asset combinations that are specifically geared for the risk and return objectives of a particular investor.

Anne Kennedy is vice president, director of corporate bond trading at Munder Capital Management in Birmingham, Michigan where she is responsible for managing institutional portfolios, trading corporate bonds and managing client relationships, focussing primarily on health care and captive insurance related portfolios.