Safe Withdrawal Rate

A classic study is the 4% safe withdrawal rate (SWR) rule in personal finance. I have done the exercise over my Historical Return Tool. It mostly align with other similar study. However, I have found some issues with the Risk Parity Asset allocation high SWR. Finally, my finding also align with the recommendation of a all equity portfolio. 

What is the Safe Withdrawal Rate?

Looking to find a single safe withdrawal rate is though. A proxy is to withdraw from the balance at retirement say 4% and then increase it by inflation each year. Running this along my historical return tool, we get the highest withdrawal rate that did not produce 0$ balance after say a consecutive 30 years  periods. 

Looking it another way, it looks at the historical worst return period and establish the best withdrawal rate that could remain viable. By definition, the SWR is a very conservative period.  Say your investment return 8% per year, withdrawing 4% allows a 2% for inflation increase + 2% of LifeStyle increase. But on the other end, Getting to 0$ in balance is disastrous. Personally, I find that using Accumulation Dynamic Decumulation is a better process to manage the withdrawal as it prudently increase or decrease the withdrawal to maximize the fund use.   

Results - All results below are based entirely on historical returns. 

Based on 30 consecutive periods (over 1945 to 2025) a 

  • 4.10% SWR : 60% Universe Bonds (Bonds) / 20% US Equity / 20% Can Equity
  • 4.40% SWR : 50% US Equity / 50% Canadian Equity 
    • 5% WR produce 10% ruin scenario (ie : no money after 30 year)
    • 6% WR produce 25% ruin scenario
    • 8% WR produce 37% ruin scenario
    • 9% WR produce 63% ruin scenario
  • 4.95% SWR : 34% US Equity / 33% Canadian Equity / 33% Develop countries Equity *over 1965 to 2025*
  • 3.70% SWR : 100% US Equity 

    Based on 50 consecutive periods (over 1945 to 2025) a 

    • 3.55% SWR : 60% Universe Bonds (Bonds) / 20% US Equity / 20% Can Equity
    • 3.95% SWR : 50% US Equity / 50% Canadian Equity 
      • 5% WR produce 10% ruin scenario (ie : no money after 30 year)
      • 6% WR produce 25% ruin scenario
      • 8% WR produce 37% ruin scenario
      • 9% WR produce 63% ruin scenario
    • 4.45% SWR : 34% US Equity / 33% Canadian Equity / 33% Develop countries Equity *over 1965 to 2025*
    • 3.45% SWR : 100% US Equity 

    While high, the correlation discrepancy does lead to a good 0.5% increase in the SWR. While if we go for a higher diversification *but with a period covering only over 1965 to 2025* including International Equity, the SWR remains 4.40%. 

    If you are looking for confirmation bias that Bonds are good in retirement, extending the Canadian Equity & Develop countries return from 1928 to 1945 with the US returns, produce the following: 
    • 4.10% SWR : 60% Universe Bonds (Bonds) / 20% US Equity / 20% Can Equity
    • 3.75% SWR : 50% US Equity / 50% Canadian Equity
    • 3.75% SWR : 34% US Equity / 33% Canadian Equity / 33% Develop countries Equity  

    Risk Parity Allocation; is it the next greatest thing?

    There`s currently drama among the FIRE community. The Clan supporting Risk Parity Asset allocation maintain that using less Equity, but more short term US Treasury and more Gold and more Commodity increase the SWR. The Blog Early Retirement Now (ERN) disagree and mentions that it is not true. 

    I have build two portfolio that nearly align with Risk Parity Portfolio (or aware according to ERN) of 20% Gold; 20% Short Term US Treasury; 20% Universe Bonds; 20% US Equity; 20% Canadian Equity (Near Golden Portfolio) and the a similar portfolio but reducing gold to 15% and adding 5% of commodities (Near Risk Parity Portfolio). 

    My data for Commodity returns from Bank of Canada only goes to 1972. Doing the analysis on my data (without extending anything) reveals that  :

    Based on 30 consecutive periods (over 1972 to 2025) the Near Risk Parity Portfolio would allow for 5.85% SWR. Based on 30 consecutive periods (over 1945 to 2025) the Near Golden Portfolio would allow for 4.50% SWR. Awesome! Seems like Risk Parity is the way to go! 

    However, if we extend the Canadian equity by making them equal to US equity before 1948 and the commodity equal to Gold before 1972, the story change drastically. 

    Based on 30 consecutive periods (over 1928 to 2025*Extended*) the Near Risk Parity Portfolio would allow for 4.10% SWR. Based on 30 consecutive periods (over 1928 to 2025*Extended*) the Near Golden Portfolio would allow for 4.10% SWR. The Risk Parity producing the exact same 4.1% results as the 60% Bonds / 20% US / 20% Canadian Equities. 

    Another concerns is that comparing with historical return the SWR usually crash around the 1970's where the equity returns were poor, inflation was high and the Gold and commodity were extraordinary good. Gold average 27 % return after inflation; while commodity did 13% on that decade while equity return 4% on average. However, the next four decade were abyssal for these. Gold average 1% and commodity -1% after inflation.  

    For me at least, Risk Parity Portfolio is only an illusion. It over relies on the 1970`s returns to remove the bad sequence using the asset class that did good during this time to boost its SWR. In many SWR study, the 1970`s were the worst period to retire, due to High inflation.  Both Gold & Commodity perform extremely good in the 1970, but I fear that the future will not comply with past pattern.

    Why gold was so good in the 1970? They literally change from a fixed USD to gold ounce before to a floating system. Gold saw more than 100% increase in 1979! Thus any portfolio containing gold allocation would look awesome for any period containing 1979. Before 1970, you could always exchange 1 ounce of gold for 35$ or the inverse. it was not a inflation protection, quite the opposite 

    • 35$ USD = 1 bar of gold = 10 loaf of bread in 1950, now say during 1950 there was a 10% inflation 
    • 35$ USD = 1 bar of gold = a little over 9 loaf of bread in 1951 

    After 1979, sure gold could start to become an inflation edge... However, did you know that 2.3 tones of gold is extracted each year . This is 1.5% of the total current extracted gold or about half of the US reserve.  Based on this alone we would expect the gold value to collapse by as much. However, people continue to want to hold gold in their asset allocation thus those who had already invested in gold increases their net worth, much like bitcoin where it price increase as more want to held bitcoin.     

    Similarly the Commodity market were also awesome in the 1970`s. Returning 13% after inflation. The 1970 were plague by a new consortium the OPEC (Petrol). I concede that higher commodity price leads to lower Equity returns. However, Commodity returns were abyssal for most of the past except the 1970`s. Will another evil consortium surge in the future to increase commodity price, perhaps but my crystal ball do not let me see that. 


    Looking at the 2000 decade, while the US equity were abyssal, the results of gold and commodity were not enough to compensate them. 

    Conclusion
    While fun to discuss the SWR, it is not the end all be all. It is merely a rule of thumb based on past historical returns. Some have tried to push it in the future using variance/covariance/expected return and such assumptions. But then you push against the reality that setting these assumptions is hard. The estimate of  financial planner in Canada for Canadian equity is 6.3% = average (CPP expect 6.65%, Expected 5.9% from DB plan, Survey among CFP 6.53%, 50 yr historical return 9.08%, Shiller cap approach 5.9%) - 0.5%. Among the prior, it is strange that the real life is so far off from the rest of the expert.  

    4% SWR seems about right on a 30 years period and a 3.5% SWR seems about right over a 50 years period. But SWR is just the start. Using model such as the Accumulation-Dynamic Decumulation method that adjust speeding when returns lower allows to get 5% withdrawal rate with reasonable down withdrawal .      

    Note: To produce these results I use my Historical Return Tool using return in Canadian Dollar. 

    Commentaires

    Posts les plus consultés de ce blog

    Historical Return Tool

    Outil: Rendement historique

    Investissement