Monday, January 23, 2012

From the mailbag

A reader writes:

I just found your blog and as an individual investor am really appreciating your perspective. I've been looking for personal finance information grounded in data and peer reviewed academic literature. I am glad I found your site. A couple of questions...

Thank you.  Your description of my blog makes me quite happy, as "personal finance information grounded in data and peer reviewed academic literature" is really a good description of what I am trying to do.

1) Has there been a version of the Trinity Study that looked at a more broadly diversified portfolio (domestic large, domestic small, international large, international small, EM, REITs, Commodities, and varying fixed income products)?

No, not that I am aware of. There have been studies adding small caps to the round up of usual suspects (large cap stocks, bonds, sometimes bills). And there have been studies which add international large cap stocks to the usual suspects too.  But nothing coming anywhere near close to your long list of asset classes. For the Trinity Study style historical simulations approach, this would basically be impossible, as there isn't enough data to do it effectively. Monte Carlo would be the only option.

But this gets to an important point. It isn't really what happened in that past that matters so much as what will happen in the future. Especially for something like EM, past may really not be prologue.

The question you ask, though, is exactly what I was seeking to provide a framework to answer in my article "Capital Market Expectations, Asset Allocation, and Safe Withdrawal Rates" from the January 2012 Journal of Financial Planning. You've got at least 8 asset classes there. I explain an approach to figure out what you might expect given your assumptions about those asset classes.

For anyone reading, if you put together a table of numbers like what you see in my Table 4 from the article for as many asset classes as you wish, and you also specify an acceptable failure probability and a desired retirement length (10,15,20,25,30,35, or 40 years), then I can produce an equivalent of the Figure 3 for you and let you know the sustainable withdrawal rate and optimal asset allocation.

This really has to be customized to each individual as there is no universal agreement on what all those asset classes will do or on how they relate to each other.

2) Have you written about how someone can develop a plan to implement the floor/upside approach when in the accumulation stages (30+ years till retirement)? There seems to be several people who are advocating saving less for retirement and smooth out consumption across a lifetime and not delay gratification to the point you won't be able to enjoy the money you have saved. With that context it seems hard to nail down what the "number" you to need with the income stream you get from annuitizing, so dependent on current interest rates. Currently I've just used the initial 4% withdrawal based on projected spending and an assumed inflation rate. I know there are quite a few assumptions involved with this approach, but spending and inflation seem less sensitive than interests rates and the subsequent annuitized income.

It sounds from your question like you are already familiar with the work of Laurence Kotlikoff.  If not, please check Spend 'til the End by him and Scott Burns. The book describes his ESPlanner software.

I do think I have something of my own to add to the answer as well.  I have not gotten to the point where I can do anything nearly as sophisticated as ESPlanner, but please see my article, "Safe Savings Rates: A New Approach to Retirement Planning over the Life Cycle" from the May 2011 Journal of Financial Planning. Be aware, though, that this article branches off from the safe withdrawal rate literature rather than the floor/upside literature.

When you are far away from retirement, I agree that it can be fine to plan in terms of a 4% withdrawal rate, but I think an even better approach for young people far away from retirement is to focus on the savings rates instead of withdrawal rates. I agree that it is very hard to focus on "The Number". 

However, it does seem that you have an entirely different reason for why it is hard to know your number, because you want to annuitize and don't know what the interest rate will be at retirement. That is a slightly different issue, since my article doesn't include annuities.  One way to deal with the specific interest rate problem for annuities is that as you get closer to retirement you can start phasing in multiple annuity purchases over time. This doesn't completely fix the problem, but it does hopefully lessen it a bit.

As your question is rightly pointing out, when you are still in the accumulation phase, you still have trouble to know it is going to be needed at some distant future date to establish your floor. There may be some other sophisticated financial engineering techniques that may help, but generally I think this is a difficult problem with no easy solution.