Tuesday, January 15, 2013

New research article: The 4% Rule is Not Safe in a Low-Yield World

I've completed a new research article along with Michael Finke and David Blanchett. It is called, "The 4% Rule is Not Safe in a Low-Yield World." The subject matter of the article will not be entirely new to readers of my blog, as I have been pushing for quite a while against the indiscriminate use of historical averages when simulating the potential success of different retirement income strategies. In the article we expound and that point, and we also look at what might happen if interest rates do go up at some point in the future. We find that with sequence of returns risk, this may not be as helpful to current retirees as seems to be implicitly assumed by many people. 

Our main points include:

The safety of a 4% initial withdrawal strategy depends on asset return assumptions. 

Using historical averages to guide simulations for failure rates for retirees spending an inflation-adjusted 4% of retirement date assets over 30 years results in an estimated failure rate of about 6% with Monte Carlo simulations.  Looking at overlapping periods from the historical data, 4% always worked.

The modest 6% failure rate rises sharply if real returns decline. 

As of January 2013, intermediate-term real interest rates are about 4% less than their historical average. Calibrating bond returns to the January 2013 real yields offered on 5-year TIPS, while maintaining the historical equity premium, causes the projected failure rate for retirement account withdrawals to jump to 57%. The 4% rule cannot be treated as a safe initial withdrawal rate in today’s low interest rate environment.

Next, it seems as though many people may wish to assume that today’s low interest rates are an aberration and that higher real interest rates will return in the medium-term horizon. Although we find little evidence to support this assumption, we estimate how a reversion to historical real yields will impact failure rates. 

Because of sequence of returns risk, portfolio withdrawals can cause the events in early retirement to have a disproportionate effect on the sustainability of an income strategy. 

We simulate failure rates if today's bond rates return to their historical average after either 5 or 10 years and find that failure rates are much higher (18% and 32%, respectively for a 50% stock allocation) than many retirees may be willing to accept. What's more, this is really a best case scenario because we do not account for any potential capital losses which investors my experience with a sudden rise in interest rates, and because we have not added any portfolio fees into the analysis.

Here is a more detailed look about how different return assumptions impact these failure rates:

The success of the 4% rule in the U.S. may be an historical anomaly.

What should one do about this? Exploring this question is really one of the fundamental purposes of this blog. Retirees may wish to consider their retirement income strategies more broadly than relying solely on systematic withdrawals from a volatile portfolio.


  1. Thanks for the recent research Wade. It may not be good news, but we need to hear it. The analysis of retirement under current economic conditions is invaluable, as is knowing that a return to better bond rates in 5-10 years won't alone be enough to save us. It seems that diversifying retirement income strategies away from just systematic withdrawals is essential.

  2. Wade
    From your prior article am I to gather that for a 40 year withdrawal period, a 20% equity allocation and withdrawals linked to the CPI that a 2.8% withdrawal rate has a failure rate of 5% or less?
    Many thanks

  3. Excellent work, Wade! Whenever I find myself with specific questions you consistently deliver the answers. I was just in conversation with a client yesterday over this exact issue. Unfortunately, I didn't have the research to back up my assertions. Now I have that research thanks to yours, Michael's, and David's excellent work. Thank you!

  4. Very scary because too few people understand the "new" reality brought about by low interest rates. Thanks for the great research and the important information you are disseminating.

  5. The last time I checked, a 65-year-old couple could not get 4% on an inflation-adjusted annuity, and that of course includes consuming the principal. So it's not surprising that one can't be confident of 4% without consuming the principal. One can probably improve the chances a bit with a somewhat different asset allocation (I presume that the insurance company selling the annuity allocates almost 100% to TIPS), but that's far from a sure thing. The same couple could get close to 4% on an annuity with a 3% yearly increase, but of course that means assuming substantial inflation risk. By the way, do you know a good source for historical annuity payouts? There is limited information on simple fixed annuities at immediateannuites.com, but I'd hoped for something longer term and more versatile in its coverage.

  6. Thank you very much everyone!

    Anonymous 1 about the 2.8% withdrawal rate: what article are you referring to? From my January 2012 article, I used simulations based on historical data, and 2.8% has a slightly more than 1% failure rate over 40 years with a 20% stock allocation. But the point of this new research is to further point out that using historical averages isn't appropriate these days. Table 3 in this article shows a much lower sustainable withdrawal rate for those circumstances.

    Anonymous 2: exactly! Though the 4% rule allows you to consume principal. But annuity companies can pool risk as those who die soon subsidize payments to those who live long. Thus they can pay based on assuming one will live to their life expectancy. Those planning on their own need to plan for living much longer. That means a lower withdrawal rate. Arguing against annuities because payouts are low compared to the 4% rule is mixing apples and oranges.

    About historical data, the best example I know about is here:


    Do note that this is Canadian data. I don't know about historical data in the US.

  7. Wade, in table 2 above, bottom two rows, how did you arrive with a 8.6% real stock return? that seems rather "lofty." Can you elaborate?


    1. Forrest, 8.6% is the historical average inflation-adjusted large-cap stock return since 1926 from the Ibbotson data. It does seem lofty, doesn't it? The point there was to illustrate how failure rates are still rather high even if these lofty historical averages return in 5-10 years.

  8. This data doesn't fit will Zvi Bodie's 100% TIPS portfolio.

    1. Rick, thanks. To clarify, I think you mean that it will be exceedingly expensive to build one's retirement around 100% TIPS. Is that right?

  9. Have you done any recent studies regarding the use of SPIAs (instead of bonds) with stocks? Would you still use 30/70 SPIA/Stocks for retirement?


    1. Hi,

      For your first question, the answer is yes. You can read about that here:


      Regarding your second question, that allocation is part of the frontier and so it is acceptable. There are lots of other possibilities too. The one you would choose depends on your personal circumstances.

      One thing to keep in mind is that these are the theoretical results. But if you do SPIAs and stocks, it means that when you are looking just at your financial portfolio, it will be 100% stocks. If that makes you uncomfortable, and you worry that you might sell your stocks after a market downturn, then it would be reasonable to move away from the efficient frontier and include some bonds too. That is less efficient, but the efficient frontier assumes you will not do any panic selling after a market downturn.

  10. Wade - thanks as always for continuing to share your research with us. Most planners I know are still using historical rates of return to estimate future returns and the 4% "rule".

    This article mentions your calculator for estimated future returns based on yield and stock performance. Could you point me to that original article and the calculator?

    Thank you!

    1. Thanks Kathy.

      The other article is this one:


      However, I've come to worry about the reliability of that approach in recent years since it requires making predictions outside the range of historical experience. There seems to be a particular problem with the dividend yield.

      I've come to think the approach outlined here will be more reliable.

      Best wishes, Wade

  11. Wade, do offer advertising at all on your site? Thanks.

  12. I'm glad that you published something about the 4% rule.