Hope everyone is having a happy holiday season. We are in our winter location for this year, San Diego, enjoying the relatively warm weather, the surfing, hiking, catching up with friends, and just having an overall relaxing time. Today I wanted to follow up on my last post on investing allocation during retirement. Here I want to take a look at the traditionally recommended approach to asset allocation in retirement and how it compares to two other approaches. The conclusion is quite remarkable and as you may suspect the traditionally recommended approach is not all its cracked up to be.

First, lets define the traditional recommendation to asset allocation during retirement. The most common approaches are generally known as life cycle models. Life cycle models basically mean that you take a more aggressive approach to riskier assets when you are younger and then transition to more conservative allocations as you get older. For those starting retirement this usually means an initial investment allocation around 60% stocks, 40% bonds. As you progress during your retirement the recommendation is to switch the allocation to more bonds, maybe 40% stocks, 60% bonds during the middle retirement years, then maybe 20% stocks, 80% bonds during the final retirement phase. Makes intuitive sense from a big picture level. So, my question as usual was does this give me the best retirement? Is this allocation the best approach? What are the trade-offs? Armed with my new personal retirement database I can now run the models and find out.

I decided to compare three different asset allocation models during a 30 year retirement using the stock and bond data going back to 1929 and running through 2011. The first asset allocation model is called the traditional model which I described above and uses an allocation of 60/40 for the first 10 years of retirement, then switches to 40/60 during years 11 through 20 of retirement, and then finishes with a 20/80 model for years 21 through 30 of the retirement period. The second model is called fixed and just uses a fixed allocation to stocks and bonds during the entire 30 year retirement period. I used the often recommended 60/40 allocation for this analysis. The last model takes the exact opposite approach to the traditional model, in this model I start with a conservative 40/60 model during years 1-10, then 60/40 during years 11-20, then 80/20 during the last 10 years. Definitely, non-traditional to say the least. Now lets look at the results of the models in terms of safe withdrawal rate (SWR), maximum drawdown (aka the sleep factor), and average end wealth.

The table below shows the results of my analysis. The most starling results is that the traditional retirement model gives the worst results across the board; the lowest SWR, the biggest drawdown, and lowest average end wealth.

Another surprising result is that the inverted model, where you get more aggressive as you get older, yields the highest SWR, the lowest drawdown. And the fixed model for many would seem to offer the best trade-offs between all three retirement parameters. So, while the traditional retirement recommendation for asset allocation seemed intuitive at the start, it turns out to give you the worst results. This is what I’ve come to expect from most retirement advice. No one seems to bother to actually run the numbers versus the historical record. A sad reflection on the state of the investment advisory business in particular for retirees. Now, most people wouldn’t be able to stomach the actions required of the inverted model but the results do bring to the fore some important characteristics of retirement. The most important is that the safe withdrawal rate is most heavily impacted by investment returns during the early years of retirement. Big drawdowns early in retirement put you in a big hole that is hard to get out of. That is why the inverted model that started at 40/60 for the first 10 years had a higher SWR than the other two. You also can’t get too conservative, for example starting at a 30/70 allocation, due to the pernicious effects of inflation. The other important takeaway is that in general you need more risky assets than you think. The traditional model had the lowest average allocation to stocks throughout the whole retirement period, while the other two models had the same stock allocation on average. This also has to do with inflation’s effect on withdrawals. Also, the average end wealth is primarily determined by how much and how long you are exposed to stocks.

In short, the traditional retirement asset allocation recommendations are not the best choices for retirement. A simple fixed allocation throughout retirement or even a model that implements the exact opposite recommendations yields better results in terms of safe withdrawal rate, drawdowns, and average end wealth.

*Full Disclaimer - Nothing on this site should ever be considered advice, research or the invitation to buy or sell securities. These are my personal opinions only.*

thanks for an interesting&concise analysis. I wonder if selecting for dividend paying stocks would affect the results.

Choosing div stocks vs the SP500 wouldn’t change the overall principle of the results, the traditional life cycle model advice is flawed, but they would obviously affect the specific numbers.

Paul

Interesting analysis. Based on what you learned, have you changed your allocation (I think I recall that in Nov you went 70-80% into cash)? If not, why?

Hi Jim, I’m mostly cash in my trading accounts. My long term focused accounts which would be applicable to this analysis, I use the IVY timing model (which is currently fully invested). In these articles I’m looking at standard retirement advice for most retirees for whom I recommend the IVY portfolio or the Permanent Portfolio.

Paul

even in my late 50’s I am still a fairly aggressive investor as it is what works best for me – I am rarely more than 25-30 percent cash.

I consider myself to be an aggresive investor. Only a small part of my portfolio is in cash and bonds.

I’ve always been suspicious of life cycle models, and it’s good to see some data refuting them. It would be interesting to see the calculations flipped to provide an SWR # for a range of portfolio mixes.

Oh Just read your previous blog and saw you already covered that…..