But is there some other option that isnt part of this analysis that is better (where better is defined based strictly on historical data with no guarantee of applicability to the future)? There is an additional asset class (only available recently) that looks very promising in the context of this kind of analysis. As I write this document Treasury Inflation Protected Securities (TIPS) are offering a real return (after adjustments for inflation) of around 4.15% and I-Bonds (savings bonds) are offering 3.4% after adjustment for inflation. Even the 3.4% inflation adjusted return is absolutely unprecedented (over the long term) in the history of the Ibbotson data of fixed income instruments. I-Bonds have guaranteed liquidity at guaranteed prices (a minor penalty applies to redemptions prior to 5 years), gains that are 100% tax deferred, and are not subject to state (or local) income taxes at redemption.
If I were to replace the intermediate government bonds class with I-Bonds and repeat the first study (see The Methodology) you would be able to increase your withdrawal target from $44,000 to $49,000 per year (note that this is an 11% increase in income) for the case of 100% portfolio success. In this assessment I made some assumptions about future availability of I-Bonds with real yields of 3.4% that may not hold true in the future.
The results using TIPS (per their valuation in late Feb. 2000) would potentially be even better if tax implications were ignored (which is quite reasonable in tax-deferred accounts and pretty much impossible in taxable accounts) and their pre-maturity open market sale price was fairly stable over the long term (not necessarily true). It is also fair to say that adding short-term and/or intermediate term corporate bonds (asset classes for which I had no data) would also (probably) be an improvement over just intermediate government bonds and T-Bills. This is NOT a guarantee that these vehicles would be an improvement over other short term investment vehicles such as investment grade short-term corporate bonds or carefully selected CD's (investment classes for which I have no long term data), but I believe that I-Bonds (for taxable accounts) and TIPS (for tax-deferred accounts) are quite competitive with conventional short-term fixed term investment vehicles.
Im not suggesting that people put half their savings into savings bonds, but this data did compel me to put a portion of my portfolio into I-Bonds and a portion into TIPS (in a 401K).
A lot more could be done. A few items that come to mind are:
1) Extending the analysis to timeframes other than 30 years
2) Other asset allocations
3) Adding I-Bonds and/or TIPS to the asset mix
4) Doing a genuine Mean Variance Optimization analysis to find the "Efficient Frontier for Withdrawals". See William Bernsteins's absolutely outstanding Efficient Frontiers site for everything that a mere mortal might need to know about Mean Variance Optimization, MPT, and the Efficient Frontier. For a less academic and complete, but more readable discussion of the same topic check out Frank Armstrong's site and follow the "Investing for Keeps"and "Investment Strategies for the 21st Century" links on his page.
5) Other withdrawal limits and algorithms, such as changing the asset allocation over time.
6) Algorithms to handle cases of delayed income streams (such as Social Security being paid after 'x' years), short term expense parameters (such as a mortgage being paid off in 'x' years), and the relative impact of expenses known to be fixed (such as mortgage payments).
But I have a real job and this stuff does take a little time. So I don't know if there will be next steps or not.