Care and feeding of User Defined Assets (UDAs)

Or: An early introduction to Asset Modules, private and shared.

As the more financially astute members of our community of customers (or as we affectionately refer to them, financial geeks) know, ESPlanner allows you to extend the assets with which you can plan. These extended assets are called "User Defined Assets" and I'll refer to them as UDAs in the rest of this note.

Unfortunately, with great power comes great responsibility.

The ESPlanner inputs for UDAs are:

1. the name,
2. the average nominal rate of return,
3. the variance in terms of the variance of large cap stocks (relative risk),
4. and the eBeta.

Beta in normal financial terms is defined in terms of the nominal rate of return. In ESPlanner it is defined in terms of the natural log of 1 + real rate of return (ln(1 + r)) and will be referred to in this note as "eBeta" standing for ESPlanner Beta. The calculations to construct eBeta are otherwise the same as those for beta. The point being that you cannot use published values of beta as a replacement for eBeta.

It is our fault that you haven't been warned about the difference between beta and eBeta. We didn't really put that together in our heads until recently.

Many of you have noticed that, occasionally, the Monte Carlo either doesn't run or provides what appear to be irrational results when using UDAs.

Larry and I have spent a couple of days working out why.

And the answer is:

As conceived, UDAs are flawed since they are currently purely synthetic assets and you can choose arbitrary values for the rate of return, variance and eBeta. Unfortunately, once you introduce more than one UDA into a portfolio, then the values of variance and eBeta matter.

Basically what happens is that if the wrong values for variance and eBeta are chosen, the mathematical model used by the Monte Carlo code breaks.

The good news is that the next release of ESPlanner will detect this error and the CE will refuse to run. The error is detectable through a failure in the correlation coefficient between the various assets in your portfolios.

This means that your UDAs must, in some sense, be based on an actual rate of return series with an actual nominal rate of return, variance and eBeta. These aren't really easy values to calculate as they require detailed data on inflation and large cap stocks as well as detailed information on the asset you wish to use.

The good news is that we've provided a tool that allows you to construct your UDAs using a series of returns (in either nominal or real terms) that you provide and it calculates the rate of return, variance and eBeta. You will find this tool by clicking My ESPlanner and clicking on the tab labeled Your Asset Module".

Detailed Instructions:

1. Go to My ESPlanner,
2. Click on the Your Asset Module tab,
3. Add an asset (the + icon adds data, the garbage can icon deletes data and the pencil icon edits data, all icons are at the bottom of the grid),
4. Add the series data for your asset. You may use actual or manufactured rates of return which may be either nominal or real.
5. Copy the data (nominal rate of return, relative risk and beta) from the assets grid to the UDA interface in ESPlanner.

All the numbers produced can be used directly by the ESPlanner user defined asset creation interface, i.e., percentages are already expressed in percentages, ratios are already calculated correctly, etc.

And you're done.

If you're wondering what the "Export" button on the page is all about, we will be providing a fully extensible "Asset Module" interface in ESPlanner. This allows anybody to build assets and plug them into ESPlanner (instructions provided once we've actually finished the implementation) directly. In the mean time, the numbers provided can be used in the existing ESPlanner user defined asset interface.

A side effect of the decision to provide Asset Modules is that there should no longer be any restriction on the number of assets usable in a portfolio since, in effect, all Asset Modules are what we now call "default" assets (Cash, Large Cap Stocks, TIPS, et al.).

You may also note that there is another new tab on your "My ESPlanner" page, "Shared Assets". Once you have created an asset, you may share that asset with the rest of the ESPlanner community, allowing any interested user to incorporate that data into their own asset modules upon export.

A couple of caveats. We're biased towards consistency. Much of the data we calculate doesn't make sense if there are no corresponding data, i.e., if we don't have inflation data we can't convert nominal interest rates to real interest rates, if we don't have large cap stock data we can't calculate relative risk or covariance properly. Thus there are periods of time where you may enter data for your assets but nothing changes in the calculations. This is not a bug, it's because our gold-standard financial data sources haven't provided us with the latest information. There are a couple of "under the hood" solutions to this problem, but our position is that we won't guess on this stuff and that you're on your own if you want to provide your own data numbers (in particular, LCS and Inflation data). We may give you slightly "bad" numbers, but all the numbers we give you will always be consistent with all the data we have and no data that we don't have.