An interesting article on safe withdrawal rates from the CFA website: Are Safe Withdrawal Rates Still Relevant in Today’s Low-Return Environment?
Unfortunately the article did not really attempt to answer the question posed by the title. It relied on historical data, which is admittedly chock full of risk events, and resulting asset/liability outcomes to support the sustainability of a 4% withdrawal rate.
In truth, the article did not really break any new ground in terms of historically sustainable withdrawal rates: the 4% rule of thumb and the 60/40 allocation split has been the subject of many a report in this area, over the years.
While a 4% withdrawal rate with 0% transaction and management costs might have been doable over the data period assessed (and I presume he is talking about a 4% rate adjusted for inflation), it does not take into consideration many of the costs of management. We need to look at after tax, after transaction and management cost rates, because it is the amount of withdrawal that needs to be met from capital (and not that provided by dividend and interest) that is the killer during sustained periods of asset price declines. What would have been interesting in this respect was not the maximum total return withdrawal (capital + income), but the maximum capital withdrawal that would have been sustained over the period of analysis.
It is worth noting that the average portfolio starting yield (equities + bonds) has been at historical lows post 2000 and that this could well have a bearing on sustainable withdrawals going forward, especially where management costs have not come down in a similar fashion:
The report in keeping with practically all others on the subject failed to address the necessary issue of frameworks in which withdrawals can be adjusted to take account of advanced market and economic cycles, especially economic cycles with significant valuation issues and structural economic risks, and vice versa for deep cyclical lows.
Clearly, modelling of withdrawals should have more sophisticated forward modelling of risks to returns, such that prospective returns are adjusted down for high markets and advanced economic cycles and up for the opposite. Likewise, we need to be able to harvest excess returns on assets and to increase withdrawals over and above the rate of inflation where actual returns, adjusted for risks to return, have exceeded expected returns.
Keeping a starting 4% withdrawal rate could leave investors assessed at market lows with unrealistically low withdrawal rates and no mechanism for adjusting withdrawals upwards or for managing the risk of adjustment.
Historical analysis is useful in terms of assessing the nature and the time frame of risk and in setting more conservative risk and return expectations. In a sense, the 4% rule is a conservative withdrawal assumption for those unable to generate more sophisticated risk/return assumptions. But it ignores the fact that risk and return and hence potential withdrawals can vary wildly for a wide number of reasons. Risk and return are not random outcomes of a black box. If they were I could understand the rationale for the broad brushstroke of the 4% rule.
I do agree that straightforward spreadsheet modelling of “expected returns” using expected returns derived from historical averages, where there is no allowance in the structure and the modelling for what many now term “sequence of returns risk” is an insufficient platform for modelling and managing risk, but I disagree with the dismissal of more deterministic risk focussed modelling. Portfolio structure is capable of being more dynamic in terms of adjusting for and managing the sequence of return risks seen at the peak of market and economic cycles.
Where I feel perhaps there is a conflict is the failure on behalf of many investment professionals to use more conservative risk and return assumptions for modelling the ability of portfolios to meet withdrawals and in providing portfolio structures focussed on managing liability risks. We need to be able to separate out the returns and risks used in stock and security management from the returns, risks and structures used to manage withdrawal risks.
Finally, I am not so enamoured of a lot of the long term hedging and derivative based products for managing liability risks. These tend to be very high cost and very inflexible vehicles that quite often also put the providers of such at significant financial risk. The last thing we need is for another round of risk transfers from the peripheral consumer to the centre of the financial system. Risk needs to be borne, or at most off loaded at the margin.
I deal with many of these issues in some of my older documents: