Saturday 2 July 2016

Retirement Withdrawal Rates vs Probability of Success

The 4% Rule gets bandied around pretty loosely (I sometime think dangerously so) in the personal finance (PF) world these days and on some of the forums people seem to believe in it almost religiously.  What I’m not sure about is if these same people have actually read the T&C’s of the 4% Rule.

Within the T&C’s there are a couple of pertinent points relevant to this post.  Firstly, it is based on US historic data which doesn’t seem to hold for the UK, a global portfolio or for many other countries for that matter and of course history is not necessarily a predictor of the future.  The other point about it is that it gives you a 96% chance of success historically.

Having debated/discussed PF topics and specifically FIRE topics with many of you over the years I’m finally (I can be a bit slow and a bit stubborn at times) starting to realise that I’m a fairly conservative creature and that I also like to go to a level of detail that probably few others would have the patience for.  These traits lead me to selecting FIRE withdrawal rates after expenses of 2.5% that hopefully will give me a 100% chance of success at planned spending even though I know I have the ability in my plans to cut back on discretionary spending in severe bear markets.

They also require me to do hours of research and write long winded posts over the selection of the ‘perfect’ bonds to equity allocation.  It was a comment in that post from The Accumulator, from the excellent and must read Monevator site, that pointed me to some research from Wade Pfau entitled What Do Market Expectations Have to Do With Safe Withdrawal Rates.  It’s definitely worth a few minutes of your time (remember it is US biased though) with the critical table being:

Withdrawal rates for an accepted failure rate and asset allocation (S&P500 and Intermediate-Term US Government Bonds)
Click to enlarge, Withdrawal rates for an accepted failure rate and asset allocation (S&P500 and Intermediate-Term US Government Bonds)

This table gives me a couple of reasons for thought.  The first one is not surprising.  If you are prepared to accept more risk of failure then your withdrawal rate increases.  However let’s put that in context.  Let’s say you want to withdraw £25,000 per annum, your investment expenses are 0.3% per annum, your investment horizon is 40 years and you want a 1% failure rate historically you would need wealth of £25,000 / (2.8% - 0.3%) = £1,000,000.  That number sounds familiar and for many, including me, it’s also a very big number.

Now instead let’s assume a 10% failure rate.  Maybe your plans allow you to cut-back in the bad times, maybe you’re prepared to go back to work of some kind if we have a severe bear market or just maybe you’re less risk averse than I am.  All of a sudden that million pounds becomes £25,000 / (3.8% - 0.3%) = £714,285.  That is a big difference when you’re slugging it out in a tough job that you don’t enjoy (or maybe never enjoyed) as much as you maybe once used to and are dreaming of FIRE.

The second food for thought is just how insensitive the withdrawal rate is to the bonds to equity ratio.  Let’s stay with the 40 year, 10% failure rate example.  For that historic maximum withdrawal rate of 3.8% you needed a 58% stocks and 42% bonds portfolio.  However drop that withdrawal rate just 0.1% to 3.7% and all of a sudden you don’t need perfection as a range of stocks between 39% and 78% would have worked just fine.

Definitely food for thought and as always DYOR. 

8 comments:

  1. Thanks RIT, very interesting table which rams home the not-always-obvious relationship between withdrawal rates, risk and timeframe.

    On asset allocation, I think it's really overcooked most of the time. Yes, diversify across asset classes by all means, but worrying about it beyond 10% increments is a waste of time. In the real investment world, uncertainty swamps detail every time.

    ReplyDelete
    Replies
    1. Hi John
      "but worrying about it beyond 10% increments is a waste of time" That's certainly what this study seems to be saying.

      Delete
  2. With these percentage success rates for a withdrawal rate of 4% or whatever, presumably, it would be considered a success if you died age 90 with £1 in savings. However, would you really want to reach age 89, having seen your £1,000,000 portfolio dwindle to £25,000 over however many years, thinking, "Oh heck, If I survive another year I'm in big trouble"

    ReplyDelete
    Replies
    1. Hi Anon
      I'm with you. It's not something I personally want which hopefully came across with my comments about my conservatism. But I'm not everyone and we all have different risk tolerances.

      The problem with draw down off investments is that you're trying to protect yourself against the worst sequence of returns. Flipping it on its head a 10% failure rate means you're likely to leave money on the table in 90% of cases. Modelling I've done and that is also in the public domain also suggests that in some cases that is going to be a very large amount of money.

      Delete
  3. I've got very similar results assuming a constant rate of return (after inflation and costs) of 1%.

    ReplyDelete
    Replies
    1. Hi Anon
      Could you explain more? The investing world doesn't return constant rates it returns a varying sequence of returns. During accrual it's less of a problem but in drawdown it can be catastrophic if you get a bad set.

      Delete
  4. Surely with your overall level of capital, you don't need a withdrawal rate. Shouldn't you be able to survive, or actually live quite handsomely, on dividends alone?

    ReplyDelete
  5. @Merceron...

    I too wonder about the relevance of (S)WR calculations when you're living off a dividend stream. Which is how I've been living for the past six months.

    At the outset, I set myself a "salary" for the year which was 80% of the divis received in the prior year. My "salary" will increase with whatever inflation / payrise rate I decide to give myself each year.

    I also have a reserve pot held as cash (or cash-like) which, with no divi or other income, would pay my "salary" for two years. In years when divis>salary (as at the outset), the reserve pot gets added to and during divi droughts when divis<salary, the reserve pot gets drawn from.

    At least, that's the plan. Ask me in 30 years whether it worked!

    While I *can* calculate a withdrawal rate, my logic doesn't stem from a withdrawal rate.

    ReplyDelete