This page presents a portfolio simulation applet that allows you to compare a ratcheting portfolio strategy I am proposing with a more traditional strategy. The same applet allows you to compare two traditional strategies or two ratcheting strategies with each other, as an aid to selecting parameters.
The portfolio simulation tool can be accessed by pressing the above button. More information follows, or see the practitioner-oriented article (200K PDF), or it is possible to access the applet and experiment with it without further ado. Because the applet is written in the Java programming language, it can probably be accessed directly in your Web browser if it current enough, whatever type of machine (Apple, IBM compatible, SUN Sparcstation, or whatever) you are using. For example, on Pentium machines, the applet seems to run well in Netscape version 3.0 or later on Windows 95 or later operating system.
The new strategy coordinates the spending rule with the portfolio allocation to avoid difficult periods when real spending from endowment is falling. To protect a level of spending, part of the portfolio is kept in a ``protected'' part that is in fixed-income assets (and in an ideal world indexed fixed-income assets that a riskless in real terms). This protected part of the portfolio is intended to maintain protected spending, which is either arranged so that spending can never fall or so that the rate of decline is limited to a constant specified in advance. The rest of the portfolio (besides the protected part) can be placed at risk, and is called the ``cushion.'' This part of the portfolio is invested in risky and riskless asset classes in fixed proportions.
For the spending rule, we ``normally'' maintain real spending (or reduce it at the maximum rate permitted by our rule), but we may increase it if fortune has smiled on our portfolio. In particular, there is some critical proportion of the endowment below which our budget cannot fall; as the endowment grows, we increase spending in our budget as required to keep from going below that critical proportion.
On the left half of the applet window, there are two graphs. The top graph shows the simulated wealth process for two alternative strategies (plotted on the screen in red and blue, but not distinguished in a greyscale printout). The bottom graph shows the spending rule for the two strategies. In both graphs, the numbers are based on an initial endowment of 100 in year 0.
On the right side applet window, there are several controls. The button on the top generates a new simulated random draw. Pressing this button repeatedly gives us a view of the variety of different simulated randome draws. Below this button, there are three different areas. The top two allow us to change the parameters for the two (red and blue) simulations. The bottom area allows us to change the parameters of the underlying stock and bond returns.
In the applet, you can compare two strategies at a time. Each strategy can be chosen to be a traditional strategy or the strategy I am proposing. The program refers to the proposed strategy as a ``ratcheting'' strategy since the property of increasing spending but never decreasing is similar to the function of a ratchet. Therefore, we can compare two traditional strategies, two ratcheting strategies, or one of each (the default). Within the parameter area for each strategy, the heading tells the color (for the plots) and what type of strategy it is, for example, ``Blue: Ratcheting Strategy.'' The button below the heading switches to the other type of strategy. Switching strategies will make one of the strategies ``disappear'' in the plot unless the two have different settings, since the one will be plotted on top of the other. The solution for this is simple; change the settings so the two strategies are different, and the two separate graphs will appear.
For a ratcheting strategy, there are four parameters to choose. The first is the initial budget, as a percentage of the endowment. This is important because, along with the fourth parameter, the maximum rate of decline, it establishes a floor for subsequent spending. The second parameter is the minimum budget as a percentage of the endowment value. The third parameter is the percentage of the cushion invested in risky assets. The final variable is the maximum permitted rate of decline in spending. This would be zero if spending is not permitted to decline, but if it is nonzero it is the maximum rate of decline.
For a traditional strategy, the parameter choice is simpler. The first is the constant percentage of the endowment to spend. This is a fraction of a weighted average of a number of years' endowment values (given in the third choice), or just the start-of-year value (if the third choice is ``none''). The second parameter choice for the traditional strategy is the percentage of the endowment to put in the risky asset.
For the asset return parameters in the bottom section, I have provided what I believe to be reasonable parameter values. A real interest rate of 3%, an risky asset expected real return of 8%, and a risky asset standard deviation of 30%. The riskfree return is less than current yields on indexed Treasuries; using a smaller number is conservative and also reflects my skepticism that current yields will persist. The expected return and standard deviation are values that are probably reasonable for a diversified portfolio of large-cap stocks, and should be increased if riskier assets are involved or diversification is incomplete (half the endowment is committed in stock of the company founded by your most important donor).
I would like to close this section with a note of warning about the simulation applet. It does not prevent you from using strange parameters. This is good because it allows you to experiment, but you should also realize the danger in drawing conclusions based on bizarre parameter values. To keep things under reasonable control, the asset return parameters in the lower box should not be changed too much, and in particular the risky asset return should be larger than the riskless return. For the investment proportions (both for the ratcheting cushion and for the traditional proportion), choose a nonnegative number that is not larger than, say, 150%. Beyond that, the investment is very aggressive and will either waste away (with high probability) or show fantastic gains (with low probability). For the ratcheting strategy, the initial budget should be no smaller than the minimum and less than the riskless rate plus the maximum rate of decline, or else it may not be possible to protect the spending that we are supposed to protect. (Note: this is approximate and ignores investment returns within a budget year, although the program accounts for this correctly.) In fact if the initial budget is approximately the riskfree rate plus the maximum decline, the strategy will be very boring because almost all the endowment will start in the protected part and will simply be invested at the riskless rate.
Some of the big differences between strategies can arise because one spends much more than the other or one has much riskier investment than the other. This makes it more difficult to perform a sharp comparison of the two strategies. Therefore, I suggest focusing on pairs of strategies (like the defaults) in which the initial spending rate and the risky asset exposure are about the same for the two strategies.
I hope you enjoy the applet! For more information about my research or on how to contact me, please see my Web page, which contains course materials, working papers, my vitae, and more.