Developing an investment policy

A question asked privately:

How is investment policy developed?  Are there any standards?

There are indeed standards.  In the old days, investment policy was developed based on the goal picked by the investor from a five-item menu:

  • Capital preservation
  • Income
  • Growth and income
  • Growth
  • Aggressive growth

An investment adviser would have a sample policy document for each goal, which would then be adjusted for the needs and wants of the particular investor (some investors would be morally opposed to investing in tobacco companies, while others could be legally prohibited from investing in South Africa, etc.)

For institutional investors, the mechanics is usually somewhat more complicated.  An investment policy appropriate for a particular investor must be developed based on the investor’s return requirements and risk tolerance.  Moreover, an institutional investor is usually expected to be able to state these parameters numerically, using volatility, probability of shortfall, or some other quantitative risk measure.  At the same time, the adviser extensively interviews the client to elicit any constraints that should be taken into account when developing the investment policy.  Generally, the constraints are grouped into five categories:

  • Time horizon
  • Liquidity requirements
  • Tax considerations
  • Legal and regulatory considerations
  • Unique circumstances

So all in all, the process of developing an investment policy generally starts from identifying the investor’s return requirements and risk tolerance.  Sometimes it turns out that the investor wants an unattainable combination of risk and return; in those situations, a reality check may have to be administered (gently, if at all possible).  Then, an asset allocation is devised and adjusted for constraints.

How exactly is an asset allocation devised?

By using statistical analysis.  There is a well-known relationship between attainable risk and attainable return known as “the efficient frontier”:


The adviser, based on the inputs provided by the investor, chooses an attainable combination of risk and return.  The choice is fed into a program called portfolio optimizer (if necessary data is available, a simple portfolio optimizer can be constructed in Excel; professionals often use specialized software such as Ibbotson Analyst; more recently, portfolio optimizers became available as online services); the optimizer, based on historical returns of different asset classes, suggests an asset allocation that is likely to achieve the desired combination or risk and return in the long run.  It is explicitly assumed that over the long haul, risk and return of major asset classes will remain what they have been in the past (it may seem like a risky assumption, but since Roger Ibbotson first advocated it in early 1970s, it proved remarkably accurate for time horizons exceeding 20 years).  The result returned by the optimizer undergoes a quick sanity check and, if it passes, gets written into the investment policy as the strategic allocation (sometimes also referred to as policy allocation).

The strategic allocation may be tactically adjusted based on market performance expectations; asset classes expected to underperform in the next few years are underweighted, while asset classes expected to outperform in the next few years are overweighted.

How are market performance expectations formed?

There is no single way of doing it; every firm does what it can.  In equities, for example, expectations can be formed using the approach devised by David Bostian several decades ago.  Bostian advocated comparing the stock market capitalization (more specifically, the capitalization of the New York Stock Exchange) to the GDP:

Time is plotted on the horizontal axis, the ratio of NYSE capitalization to GDP, on the vertical axis.  If the ratio is low compared to the historical averages (i.e., stocks are cheap), equities are overweighted; if it is high (i.e., stocks are expensive), equities are underweighted.

A possible alternative is to use “the Fed model” (it was developed by Edward Yardeni, who never worked for the Federal Reserve, but named it “the Fed model” because Alan Greenspan mentioned a similar approach to stock market valuation in one of his speeches).  This model is based on comparing the aggregate earnings yield (also known as the Earnings/Price ratio, the inverse of the Price/Earnings ratio) to the 10-year Treasury bond yield:

If E/P is substantially above the bond yield (i.e., stocks are cheap), equities are overweighted; if E/P is substantially below the bond yield (i.e., stocks are expensive), equities are underweighted.

The fixed income industry has its own methods.  Some firms try to forecast the yield curve using structural or non-structural econometrics, others attempt to model credit spreads based on stock market volatility, but those are just two of great many approached used…

Leave a Reply

Your email address will not be published. Required fields are marked *