OBJECTIVE

“Determinants of growth in distribution portfolios” is eligible for CE credits, see Accreditation details for more information

Background

The first serious research on determinants of portfolio growth was a study by Gary P. Brinson, Randolph L. Hood, and Gilbert L. Beebower in 1984. They analyzed data over a 10-year period, beginning in 1974, from 91 large corporate pension plans with assets of at least $100 million. It’s usually referred to as the “Brinson study.”

They later expanded their analysis to include an additional 10 years of data. The new study, “Determinants of Portfolio Performance II,” was published in the Financial Analysts Journal (January/February 1995).

They concluded that the components of a portfolio’s difference in success are: asset allocation, 93.6%; security selection, 2.5%; market timing, 1.7%; other, 2.2%. Many in the financial industry have since believed asset allocation is the holy grail of investing.

When a new account is opened, the first thing a client does is fill out a risk-assessment questionnaire. Based on their answers, he or she is then pigeonholed into one of four or five investment portfolios.
This course challenges that approach.

The findings of the Brinson study cannot be applied to individual retirement portfolios for the following reasons:

  • The dynamics of cash flow in a pension fund are entirely different from those in an individual retirement account. A pension fund has a continuous inflow of money over time. In an individual retirement account, inflows occur mainly during working years. After retirement, there are usually no more inflows – only outflows. Pension funds are “open-perpetual” systems; individual retirement accounts are “closed-finite” systems.
  • When there’s a shortfall in a pension fund, contributions are increased to cover it. There’s usually no such opportunity with individual retirement savings.
  • Once withdrawals from an individual retirement account begin, the adverse effect of “reverse dollar-cost-averaging” becomes important. In a pension fund this effect is insignificant because there’s a continuous inflow of money.
  • Inflation is important for individual accounts. Withdrawals must be increased over time to maintain the same purchasing power. In pension funds, there’s no such concern: as inflation rises, salaries and pension contributions increase. Also, many pension funds have limits on how retirement payments are indexed. Individual retirees holding their own saving accounts do not have that choice—their expenses must be met.
  • The Brinson study’s timeframe is 20 years and covers only one secular bullish trend—arguably the “luckiest” 20 years of the twentieth century. Such a short time frame misses significant adverse events that occur exclusively in secular bear or secular sideways market trends.

Figure 1: Time period covered in the Brinson study

The Brinson study is still valuable, as there’s no doubt asset allocation is important for a pension fund’s success. Asset allocation reduces volatility of returns to an acceptable level; the question is whether volatility of returns is the most important factor. The answer is a resounding “No.”

Sequence of returns and inflation have a far greater impact than volatility of returns. Our analysis also uses other factors not included in the Brinson study, including:

  • Withdrawal rates
  • Rebalancing frequency
  • Portfolio costs
  • Potential alpha produced by better management of investments

Figure 2: Determinants of a portfolio’s growth

Sequence of returns and inflation are outside the control of the investor or advisor. We call them the “luck factor.” We’ll refer to the others as “manageable factors.”

Methodology

We use market data from 1900 through the end of 2010, and no Monte Carlo simulators. “Aftcasting” is our term for actual market history. It displays, in one chart, the outcome of all historical asset values of all portfolios since 1900, as if a scenario starts in each one of the years between 1900 and 2000. It provides a bird’s-eye view of all outcomes, and success and failure statistics with exact historical accuracy. This precision stems from inclusion of historical equity performance, inflation rates, and interest rates (see “Data Sources”), as well as historical sequencing of all these data sets.

Consider an example. Bob, 65, is retiring and plans on withdrawing $15,000, indexed annually to inflation, until age 95. His primary concern is sustainability of his income stream. He has $350,000 in his investment portfolio and an asset mix of 40% equities and 60% fixed income.

Figure 3 shows the aftcast for this scenario. The thin, black lines represent aftcasts. There is one line starting at the left vertical axis for each year since 1900. We define the bottom decile (bottom 10%) of all outcomes as “unlucky,” the top decile (top 10%) as “lucky.” The blue line shows the median outcome where half of the scenarios are better and half worse. In this example, the probability of depletion by age 95 is 34%.

Figure 3: The aftcast of fixed $15,000 annual withdrawals, indexed to inflation, from an investment portfolio, starting capital of $350,000

Aftcasting only shows what would have happened for a specific scenario—we do not use it to make predictions. We fully agree with those who say that past events will not necessarily be repeated in the future.

We’re interested in the frequency, size and the persistency of extreme events that happened in the past. These extremes are our starting point for designing a robust retirement plan for our clients.

Let’s look at each factor in more detail.

Sequence of Returns

The sequence of returns is the direction and persistency of the volatility of returns. Mathematically, it can be defined as the first time-derivative of the volatility of returns.

Volatility of returns by itself does not necessarily cause damage to a distribution portfolio. It is the sequence of returns that can do irreparable damage. For example, someone retiring in 1987 would have experienced a significant volatility of returns—a 30% drop in the equity index during October and November of that year.

See all comments Recent Comments
Add a comment

Have your say on this topic! Comments are moderated and may be edited or removed by
site admin as per our Comment Policy. Thanks!