Developing A Personal Investment Strategy

Developing a Personal Investment Strategy–A Practical Application

by Randy Von Fumetti

When formulating a personal investment strategy, the application of a disciplined methodology can lead to more rational decision making by the investor.

As a practicing actuary for 23 years, and chief financial officer of two major life insurance companies, I spent a considerable amount of time performing asset and liability management studies on the investment portfolios backing company annuity and life insurance liabilities. Now, as a financial planner, I develop investment strategies for personal investment portfolios. This article will discuss the development and implementation of such a portfolio from the viewpoint of a practitioner.

Actuaries make careers out of managing enterprise risk. It is a broad concept and encompasses many facets. In the context of a personal investment portfolio, risk is a much narrower concept. However, it is crucial that investors have a basic understanding of the nature of risk in their portfolios and its implications on their ability to meet their financial planning objectives. Consequently, we will begin with a discussion of risk as an introduction into the development of a personal investment portfolio.

Nature and Implications of Portfolio Risk
The concept of risk is not well understood by the average investor. Because many investors associate risk with the notion of receiving back something less than the full value of their original investment, they often exhibit more conservative investment behavior than may be optimal. In the context of portfolio development, however, risk is defined as the volatility of monthly or annual returns and is measured in standard deviation. Once investors understand how portfolio risk, or volatility, differs from the idea of receiving something less than full return of principal, they become more willing to consider alternative risk profiles.

Risk can and should be viewed differently during the portfolio's accumulation phase and distribution phase. During the accumulation phase, volatility doesn't translate into losses, but rather reduces wealth accumulation potential. The following hypothetical example can be used to demonstrate this concept to the investor. Assume two portfolios each have a beginning value of $500,000. Portfolio 1 is invested to replicate the Russell 1000 Growth Index over the twenty–year period from 1984–20031. Portfolio 2 is invested in a hypothetical investment for the same period that is assumed to earn a level return each year (i.e., no volatility) equal to the arithmetic average annual return of Portfolio 1.

Table 1 shows that the geometric average annual return is over 200 basis points higher in the portfolio with zero volatility. The additional return translates into 45 percent more wealth at the end of the twenty years. Although an extreme example, it clearly demonstrates the impact of volatility on portfolio performance during the accumulation phase.

From the investor's viewpoint, the mindset regarding risk shifts from concern over losing original principal to the need to determine an acceptable level of fluctuation in annual returns. Once educated in this fashion, it has been my experience that the investor will often take on a more aggressive risk posture, thus increasing long–term growth potential.

As a portfolio moves into the distribution phase, the effects of risk are quite different. Here volatility can translate into losses, if withdrawals are required to meet cash–flow needs in years where values are depressed due to poor performance. From a conceptual standpoint, this is similar to asset and liability management applied to an insurer's block of business. Monte Carlo is particularly helpful in demonstrating the impact that different levels of risk can have on the portfolio's potential to meet cash–flow objectives.

The level of risk in a portfolio can be demonstrated to an investor by one or more of the following means:

  • Statistically, as in standard deviation.
  • As an expression of the frequency that annual returns will vary within a range around the expected return.
  • Positioning along the risk and return spectrum, or efficient frontier.
  • Using Monte Carlo to express a variety of probabilistic outcomes.

Let's now turn our attention to the process of developing a personal investment strategy. It consists of three components:

  1. Identify constraints and parameters.
  2. Design the portfolio.
  3. Implement the design.

Identify Constraints and Parameters
Before any analytical work can be performed, the unique constraints that will affect the portfolio design must be identified and quantified. They represent the investor's goals and objectives and include:

  • Risk tolerance.
  • Performance expectations.
  • Cash–flow requirements.
  • Wealth accumulation goals.
  • Tax status.

Risk tolerance
Risk tolerance represents the amount of risk that an investor would find acceptable. The need for assessing risk tolerance is clear. The higher the risk tolerance, the more aggressively the portfolio can be invested and the higher the expected rate of return.

Risk tolerance is typically assessed by asking the client a series of questions about his attitude and expected behavior in regards to the economy, the markets and the performance of a hypothetical portfolio. Investment time horizon and age are also commonly used criteria. The client is then positioned somewhere along the risk spectrum ranging from conservative to aggressive. Another method involves assessing the client's level of comfort with the expectation that the portfolio will experience a negative annual return X percent of the time, or with the probability that the portfolio will have a cumulative loss after three years of Y percent. From practical experience, I have found discussions around such issues as why a particular investment was acquired, or the extent to which the client worries about losing money can be very useful in assessing risk tolerance.

Assessing risk tolerance in this fashion is as much of an art as it is science. Therefore, it is only one factor among many that should be considered in the development of an optimal portfolio. Risk, and its impact on the portfolio's ability to fund cash–flow needs and wealth accumulation goals, become easier for the investor to visualize when presented in the context of the efficient frontier or Monte Carlo analysis.

Performance expectations
The client may have a specific target in mind, for example, to average 7 percent over the next 10 years. In this case, the level of risk associated with a portfolio that is expected to earn 7 percent needs to be demonstrated by applying any of the methods discussed above. In most instances, the client will not have a specific return objective. The performance goals will then be established in the design phase by comparing risk and return profiles of alternative investment strategies. It is important to identify the client's objectives early in the process.

Cash–flow requirements
We typically think of cash–flow requirements as outflows only, but inflows should be considered as well. Expecting a significant infusion from the sale of a business, for example, could impact the decision today to leave shelf space in a taxable account in order to accommodate the subsequent infusion. Other cash–flow considerations include:

  • Ongoing living expenses.
  • College education costs.
  • Anticipated support for parents or other family members.
  • Business purchase.
  • Significant one time costs, such as weddings, travel, etc.

If the investment strategy is being developed in the context of a comprehensive financial planning process, detailed cash–flow projections will be readily available. They will be used in the design phase as a necessary ingredient in Monte Carlo analysis. In the implementation phase, cash–flow requirements will influence appropriate investment choices. For example, it may be desirable to utilize income–producing assets to meet certain cash–flow objectives.

Wealth accumulation goals
The client may have a specific goal in mind, perhaps to have a certain amount of funds
available for retirement or to preserve a legacy. If no specific goal is evident, modeling in the design phase will demonstrate the impact of various investment strategies on wealth accumulation potential.

Tax status
Current and projected income tax brackets, as well as the split between taxable, tax–deferred and tax–exempt accounts, are important considerations in the implementation phase. They will influence the selection of specific investments and affect the location of investments within the various account types.

Design the Portfolio
Here the principles of modern portfolio theory are used to develop an asset allocation that helps meet the client's risk and return goals. Asset allocation is the decision of how to diversify resources among a variety of broad asset classes. Harry Markowitz and Bill Sharpe demonstrated that a higher potential rate of return with a reduction in risk could be achieved by adding asset classes to the portfolio. Asset allocation seeks to maximize the potential return for a given level of risk.

Brinson, Singer and Beebower conducted a series of studies from 1986–1991 which led them to conclude that over 90 percent of long–term investment return is attributable to the asset allocation decision.2 According to their studies, security selection, market timing and other factors account for the remainder.

There are two types of risk associated with any security. Systematic (or non–diversifiable) risk is the risk associated with investing in a particular market. All investments in that market will be affected by this risk. Asset allocation cannot reduce this risk. Unsystematic (or diversifiable) risk is the risk associated with investing in a particular security. This risk is the type that we seek to reduce or eliminate through proper diversification, or asset allocation.3

There are four major assumptions needed to develop an asset allocation model:

  1. Determination of asset classes.
  2. Expected mean returns for each asset class.
  3. Standard deviation of each asset class.
  4. Correlation among the asset classes.

Typical asset classes include cash and equivalents, short–term and long–term bonds, large–cap domestic equities, small–cap and mid–cap domestic equities, and developed market international equities. A more diverse array of asset classes might also include any of the following: high yield bonds, real estate, developing market international equities, hedge funds, inflation protection securities or others. If the asset allocation model utilizes too few distinct asset classes, then diversification benefits will be limited. On the other hand, there is a practical limit on the maximum number of distinct asset classes that can be considered. For example, minimum required investment in the hedge fund universe makes this asset class appropriate for larger portfolios only.

Expected return, volatility and correlation for each asset class are key assumptions in the asset allocation model. Each asset class is represented by a benchmark. For example, the U.S. large–cap universe may be represented by the Russell 1000 index. The portfolio's return is the weighted average expected return of the asset classes represented in the portfolio. The volatility of the portfolio is then calculated as follows: (See Formula)

Many asset allocation models use only historical return data to extrapolate future return assumptions. The historical returns do not take the current economic environment into consideration. In addition, they make implicit assumptions about key economic drivers of asset returns. Examples of such drivers, or supply variables, for equities include inflation, dividend yields, real earnings growth and the impact of changes in valuation. More sophisticated asset allocation models take supply variables as well as historical data into account when establishing expected return assumptions.4

The asset allocation model then calculates portfolio returns and standard deviations over a broad array of asset class weightings. When plotted on a risk versus return graph, the results produce the efficient frontier. The frontier is an accumulation of optimal efficient portfolios. No portfolios consisting of indexes exist above the efficient frontier line and only inefficient portfolios exist below the line. For every portfolio below the line, there exists a more efficient portfolio with either less risk for the same return (Portfolio A), higher return for the same risk (Portfolio B) or a combination of less risk and higher return (Portfolio C) (see Figure 1).

How, then, is the asset allocation model applied in practice to develop an asset class mix that is appropriate for an individual investor? Risk tolerance can be plotted on the efficient frontier that will correspond to an optimal asset mix with an expected level of return. The risk and return profile is represented by an annual expected return and standard deviation. The investor’s comfort level with the optimal asset mix can then be evaluated, addressing such issues as:

  • Compared to his current portfolio, is the
    new asset mix more or less risky? How does the investor feel about that?
  • How does the expected return compare to his current return? To his desired return objective? What amount of incremental wealth creation does the difference imply over a five–year, 10–year or 20–year period?
  • Are there any asset classes in the suggested asset class mix that the investor is not comfortable holding?
  • Is the new asset mix capable of generating sufficient yield and/or liquidity to fund projected cash–flow needs?

Monte Carlo analysis is extremely useful in helping the investor assess the appropriateness of potential asset class mixes. It should ideally be run on a small set of potential asset class mixes, including the current portfolio. Comparing the current portfolio to various alternatives will demonstrate the relationship between return, risk and the ability of each asset mix to meet cash–flow objectives. The following metrics can be used to compare various strategies:

  • Probability of cumulative loss after X years.
  • Probability of meeting cash–flow needs.
  • Probability of meeting cash–flow needs and
    accumulating a specified amount of wealth.
  • Dispersion of possible outcomes.
  • Best and worst case scenarios.

Implement the Design
The design process determines the percentages of each asset class that will be represented in the portfolio. Regardless of the degree of sophistication in the asset allocation model, the portfolio will fall short of accomplishing the investor’s desired objectives if it is not properly implemented. The goal of any actively managed portfolio implementation should be to meet or exceed the risk and return expectations that have been established in the design phase. Implementation involves asset class location (versus asset allocation) and asset selection.

Asset location
Asset location identifies accounts that are optimal placeholders for each asset class. This is important in order to maximize tax efficiency and distribution flexibility. Larger portfolios are usually spread over a number of accounts, including IRAs, 401(k)s and a variety of nonqualified accounts. In addition, both spouses may maintain separate accounts. A general guideline is to place less tax–efficient assets in tax–deferred or tax–exempt accounts and more tax–efficient investments in taxable accounts. This placement serves to defer current ordinary income and allows taxable accounts to be subject to lower capital gains rates. For example, certain bonds and real estate would be preferred in tax–deferred or tax–exempt accounts. Other bonds should be held in taxable accounts, especially for high tax bracket investors. Securities that have returns linked to an inflation index are capable of generating significant phantom income during inflationary environments and should ideally be held in tax–deferred or tax–exempt accounts.

Other issues to consider include account size, liquidity needs and time horizon for each account. A practical limitation is having enough room in various account types to accommodate the ideal asset location. For example, there may not be enough room in tax–deferred and tax–exempt accounts to house the less tax–efficient assets. In this case, one option would be to consider a strategy that creates additional shelf space for the less tax–efficient assets.

Asset selection
Once the asset class and account structure has been determined, investments must be chosen. The importance of using specific criteria to select assets is necessary to ensure that the diversification benefits of the asset allocation model are achieved, and should not be understated. Specific asset selection criteria include style specificity, beta, volatility, alpha, total return, expense ratio, manager tenure, turnover ratio, tax–efficiency ratio, yield and others.

Asset selection criteria apply differently to various types of investments. For example, expense ratios are an important consideration with certain investments, but not others. Due to their popularity, I will discuss asset selection criteria in regards to actively managed equity investments.

Style specificity refers to the degree that the investment represents a single asset class, a growth or value orientation, and a capitalization level. For larger portfolios, highly style specific investments have four advantages:

  • Facilitates portfolio rebalancing. The actual asset mix will deviate from the target asset mix over time as certain asset classes outperform while others under perform. Periodic rebalancing will help to maintain the desired asset allocation. From a practical standpoint, it is much more difficult to rebalance investments that represent multiple asset classes, a combination of both growth and value type companies, or companies varying greatly in size. Furthermore, rebalancing among investments that represent multiple asset classes can trigger unnecessary capital gains taxes.
  • Improves tax management capabilities. As discussed previously, locating asset classes in the proper accounts can significantly reduce income taxes. The effectiveness of this strategy is greatly diminished if multiple asset classes are represented by a single investment. For example, an investment consisting of both bonds and stocks cannot be readily split among tax–deferred and taxable accounts.
  • Enhances ability to monitor style drift. Many investments drift from their stated objectives over time, sometimes due to pressure to generate higher returns. Significant drift signals that the investment may no longer be appropriate for the portfolio.
  • Leads to a high degree of correlation with the asset class benchmark. The degree of correlation with the asset class benchmark is probably the most important criteria in assuring that the desired asset mix is truly achieved. Ideally, the benchmarks should be the same as those that were used to represent the asset classes in the asset allocation model. Unfortunately, correlation is often one of the most ignored criteria in the asset selection process. Consequently, many investments are selected based primarily on historical returns. A lack of correlation with the asset class benchmark indicates a high likelihood that the investment won’t perform in a manner consistent with the asset class.

Correlation measures the degree of style specificity, and can be assessed in two ways. Holdings–based analysis evaluates the breakdown of an investment by asset class, growth versus value orientation, capitalization, sector exposure, cash holdings, exposure to geographical markets and others. Comparing these breakdowns to similar breakdowns for the benchmark allow for a subjective evaluation of their correlation. Returns–based analysis calculates the r–squared coefficient for the investment and the benchmark. R–squared measures the strength of the linear relationship between historical investment returns and benchmark returns. Ideally, the investment should exhibit a high degree of correlation with the benchmark under both methods.

Beta measures an investment’s sensitivity to movements in the benchmark. Thus, it is a measure of relative risk. Only with a high r–squared can we be confident that beta is credible. A beta greater than one indicates greater sensitivity to the benchmark. In this situation, the investment performs better than the benchmark in up markets and worse in down markets. Beta can provide insight to the appropriateness of a particular investment to represent a given asset class.

Volatility, measured in standard deviation, depicts how widely returns vary over time, without regard to a benchmark. It is a measure of absolute risk. The goal is to select investments with less volatility than the benchmark. A high beta does not necessarily imply a high level of volatility. For example, a specialty fund that invests primarily in gold might fluctuate widely because of rapid changes in gold prices. Thus it will have high volatility. But its price may not move as high as the benchmark in up markets or as low in down markets. Thus, relative to an appropriate gold benchmark, it may have a low beta.

Alpha measures an investment’s excess return over its expected performance, given its relative level of risk, or beta. A positive alpha indicates the investment has performed better than its beta would predict. As with beta, a high r–squared lends credibility to the use of alpha. Alpha can be used to directly measure the value added or subtracted by an investment’s active management.

Total return reflects both appreciation in net asset value and interest, dividends and capital gains distributions received. The goal is to select investments that have the potential to consistently outperform the benchmark. While total return is obviously an important consideration in investment selection, it should be appropriately evaluated in the context of all relevant criteria. Other factors to consider include expense ratio, manager tenure, turnover ratio, tax–efficiency ratio and yield.

These criteria should be rigorously applied at the time of portfolio development. In addition, they should be monitored regularly over time and changes to existing investments should be made when they no longer meet the established criteria. In so doing, there will be a greater likelihood that the risk and return expectations developed in the asset allocation model will be exceeded.

The development of a personal investment strategy should begin with educating the client on the nature of risk and its implications on financial planning objectives. Subsequent steps in the process include identification of constraints and parameters, design and implementation. Application of actuarial and finance principles throughout this process can lead to an improved risk and return profile. In addition, a disciplined methodology tends to remove the emotion from investing which leads to more rational decision making by the investor.

Randy Von Fumetti, FSA, MAAA, is a financial planner with MGC Financial, Inc., and is an investment advisor representative with Lincoln Financial Advisors, a registered investment advisor. He can be reached at: (515) 243–3212.

1 It is not possible to invest directly in any index. The performance of an unmanaged index is not indicative of the performance of any particular investment.

2 Brinson, Gary P., Brian D. Singer and Gilbert L. Beebower. 1991. Determinants of Portfolio Performance II: An Update. Financial Analysts Journal 42(4): 39–44.

3 Diversification does not eliminate risk, does not guarantee a profitable investment return and does not guarantee against a loss.

4 Historical data deals with timing risk and neglects the future risks of investment risk, economic risk, currency risk and reinvestment risk. It is not possible to directly invest in an index and to extrapolate from such historical data should only be used as a guideline. Past performance does not guarantee future results.