Thursday, June 21. 2012
The inspiration for today's blog post was a recent question I had while presenting on the topic of "Expanding the Framework of Safe Withdrawal Rates" at the FPA NorCal conference several weeks ago. While discussing some of the history of the so-called "4% rule" and where it came from, an audience member asked if the safe withdrawal rate should really be 3%, given how increasingly common it is for wealth management firms to charge a 1% assets-under-management fee. "Not at all," I replied, "in fact, the adjustment to safe withdrawal rates for investment costs is barely more than 1/3rd of the 1% you charge!"
Adjusting Safe Withdrawal Rates For Investment Costs
Over the past decade, several studies have examined the impact of investment expenses on safe withdrawal rates, given that the original research on the 4% rule did assume the investor earns index returns - unadjusted adjusted for expenses.
The first study to evaluate this issue directly was "Adjusting Withdrawal Rates for taxes and Expenses" in the April 2001 issue of the Journal of Financial Planning. Pye's results generally found that a 1% expenses ratio translated into a safe withdrawal rate decline of about 0.5%; for example, the 4% safe withdrawal rate using index returns becomes a 3.5% withdrawal rate after a 1% expense ratio/investment advisor fee.
The February 2010 issue of The Kitces Report, "Investment Costs, Taxes, and the Safe Withdrawal Rate" also examined this issue, and similarly found that a 1% increase in investment expenses reduced the safe withdrawal rate by approximately 0.4%, or about 40% of the gross fee. The Kitces research also found that this relationship tended to hold at higher and lower expense levels as well; thus, the impact of a 0.25% expense ratio was about 0.1% off the safe withdrawal rate (where 0.1% is 40% of 0.25%), and the impact of a 2% expense ratio was about 0.8% off the safe withdrawal rate.
The results of the Pye and Kitces research studies were somewhat surprising to many planners. After all, if the client is spending $40,000 from a $1,000,000 portfolio, and has to pay a $10,000 fee, how can the client's net available spending not be reduced to $30,000!?
Solving The Investment Cost Puzzle
This puzzle - how a $10,000 fee doesn't reduce sustainable spending by $10,000 - is answered by looking at how investment costs impact safe withdrawal rates not just in the initial year, but over time. For instance, the chart below shows how a client's portfolio, withdrawals, and investment costs would have varied across a 30-year retirement that started in 1966 (one of the worst retirement years in history).
The blue line is the account value (scaled to the left axis), which starts at $1,000,000, and due to the ongoing withdrawals and difficult investment environment, struggles to ever materially grow, eventually spending down rapidly in the final years. The green line (scaled to the right axis) represents the amount of the client's spending withdrawals (not including the fee itself), starting at $40,000 (a 4% initial withdrawal rate on the $1,000,000 starting account balance), and rises each year for inflation, rising to a final year withdrawal of over $188,000 just to maintain the same real level of spending after the cumulative impact of inflation (especially harsh through the 1970s).
The red line is the client's investment fee, 1% of the account balance (blue line). In light of the fact that the blue line struggles to grow - since this is a particularly unfavorable investment/spending scenario - the red line too struggles to grow, remaining relatively flat for the first 22 years, before finally tumbling in the last 8 years (along with the account value).
What the chart reveals is that in scenarios like this - where the portfolio struggles, and taking "only" a 4% withdrawal rate matters - the fee relative to the withdrawals has a diminishing impact over time. For instance, while in year 1 the fee was $10,000 and the withdrawal was $40,000 (the fee was 25% of the withdrawal), by year 20 the inflation-adjusted withdrawal was up to almost $140,000 while the fee was still only about $10,000 (the fee was only 7% of the withdrawal); in the final year, the withdrawal was for about $188,000 (all the money left) while the fee was only $1,880 (1% of the remaining account balance and also 1% of the remaining withdrawal). Because the red line (fees) declines with the portfolio while the green line (spending withdrawals) rises with inflation, the fees have less and less of an impact while the outcome becomes dominated by the impact of spending itself.
Applications and Implications for Clients
In terms of setting an appropriate spending policy for clients, the implications of the expense research are that expenses absolutely do matter, but that it's an overstatement to suggest that spending needs to be reduced by the total amount of investment costs. Instead, what the research shows is that the best policy is to simply let the portfolio pay its own costs and fees, and adjust the safe withdrawal rate downwards proportionate to the fee - which then sets the client's spending level. Thus, for example, if the investment fee is 1%, the safe withdrawal rate might be reduced to 3.6% from 4%, and from that point forward the portfolio will pay its 1% fee ($10,000 in year 1, adjusting each year based on the account balance), and the withdrawals will be $36,000/year, adjusting for inflation. Of course, if the fee is 2%, the safe withdrawal rate adjustment would be 0.8%; any particular client's safe withdrawal rate adjustment can be adapted to their own individual level of investment costs.
Notably, if the client does not retire in such a wretched scenario as the 1966 retiree, the portfolio balance will likely rise over time, which means in turn the fee too will rise. Nonetheless, this is still not problematic from the safe withdrawal rate perspective; after all, if returns are actually good - or merely "not horrible" - then the safe withdrawal rate itself would be higher. Thus, for example, a 1% expense ratio might have a 0.7% safe withdrawal rate impact in a rising market scenario, but since the safe withdrawal rate itself in a rising market scenario might be 6%, the reduction to 5.3% (adjusted for fees) still means a 3.6% safe withdrawal rate (as a worst case baseline) is still safe, and in fact likely to allow for spending increases in the future as the floor ratchets higher.
Of course, in reality this is just one of many factors that can be used to adjust safe withdrawal rates; we've previously noted on this blog how to adjust the safe withdrawal rate based on the client's time horizon as well for those who will have a retirement time horizon much shorter or longer than the standard 30-year assumption. Other factors, as discussed in depth in the recent March 2012 issue of The Kitces Report, include taxes, diversification, spending flexibility, risk tolerance, the valuation environment, and more - all of which can be incorporated to customize the safe withdrawal rate for a specific client scenario.
But the bottom line is that, at least from the perspective of investment costs, it's important to remember that the impact at the beginning is not the same as the impact for life. The good news about investment costs like expense ratios and assets-under-management fees is that, because they continuously recalculate on the current account balance, become self-mitigating by declining in spend-down situations and only rising in favorable return environments... where the clients can afford the fees anyway.
So what do you think? Do you make adjustments to the safe withdrawal rate for investment costs? Have you been adjusting too little? Too much? What other factors do you look at when adjusting spending recommendations for clients?
(Editor's Note: This article was featured in the Carnival of Retirement - 25th Edition on I Am 1 Percent.)
I took a look at this issue at my blog just the other day too. I agree with your conclusions.
I'm surprised that you didn't make any mention about how an advisor could earn that fee by helping the client get settled into an appropriate asset allocation and then stay the course. There is more to the story than just emphasizing that a 1% fee does not reduce the sustainable withdrawal rate by 1%.
I made two tables. The first is about the point you are making here, but you might be interested to see the actual numbers: I show the withdrawal rate with and without a 1% fee for each rolling period from history.
The second table I think is also interesting. I show how much lower wealth accumulations are after 30 years when someone uses a 4% withdrawal rate with a 50/50 asset allocation. I think that is also a good way to see the impact of fees.
Best wishes, Wade
I've been noticing this before, and I am seeing it now with your blog as well. Discussions about fees really doesn't attract much interest in the ways of comments, etc.
Sometimes the things I think will be hot topics (such as fees) fall flat, and then other things I just do as an aside end up getting a lot of interest. I can't figure out how to predict this.
To me it is a huge part of why low fee ETFs have been on a meteoric rise.
Thank you for your interesting posts. I think the fee discussion, particularly the advisor's fee should focus on savings a client receives by taking our advice. The Dalbar studies show this savings very clearly when comparing investment returns against investor returns. Perhaps analysis can be done demonstrating the impact of using average investor returns and a safe withdrawal rate compared to taking our advice and obtaining long term investment returns with a safe withdrawal.
You continue to blaze trails on safe withdrawal rates. On the impact to fees, however, I take issue with a few of your observations. I don’t do so lightly because I know of your skills and intelligence. Having Wade Pfau back you up, leaves me even more cautious in my comments, as the two of you represent the elite thinkers on this topic. Nevertheless, while I agree with your general thrust, I feel you have allowed math to trump common sense.
You are correct that a 1% fee against a 4% withdrawal rate is 25% of the amount withdrawn in the initial year and a smaller percentage each year thereafter. You also are correct in your observation that the decline is due to withdrawals being increased each year for inflation whereas the fees are not. So far - so good.
Where you fall of the rails is your assumption that spending remains inflated each year even when portfolio performance is dismal. We all know that is unrealistic even though it allows you to make your point. Let’s get out the box for a minute and apply some additional math. A $1,000,000 portfolio with a 4% withdrawal rate, 1% in fees and 3% inflation starts out with that 25% fee to withdrawal ratio. To understand solely the impact of inflation on the withdrawal rate, I assume the portfolio always stays at $1,000,000 thus eliminating changing withdrawals occasioned by changing portfolio values. In other words, I have isolated inflation.
The correct impact from 3% annual inflation to the safe withdrawal percentage is to reduce the 25% fee to withdrawal ratio to 22.21% after 5 years, 19.16% after 10 years and 14.26% after 20 years (contrasted with 7% in your example). These percentages are simply the inflated withdrawal rate against a constant $1,000,000 investment. These are the only percentages that matter assuming you are maintaining the integrity of a SAFE withdrawal rate.
Allowing withdrawal rates to increase by inflation without considering the value of the portfolio is akin to retirement suicide. In your example, as the portfolio starts to decline, the responsible advisor would suggest a reduction in spending, rather than the ever increasing withdrawal rate you posit. In other words, the actual withdrawal amount would be much lower if it were to remain SAFE. As a result, the fees as a percent of the withdrawal rate twenty years hence would be much higher than your conclusions if you maintained the goal of SAFE withdrawal rates.
As you know Michael, I have been a frequent advocate of low fees, in part because I believe advisors owe a duty to the public trust (that is the CPA in me). High fees as a percentage of the portfolio convey the opposite of public trust and are destined to destroy hundreds of thousands of retirements. I also suggest that all cost must be considered not just advisor fees when doing this analysis. Let’s not forget fund expense ratios, sales commissions, costs of guarantees, insurance costs of annuities and the often forgotten trading costs and spreads. When all costs are included, I have seen total costs approach 4% on variable annuities with living benefits, and by anyone’s math that is 100% of the initial safe withdrawal rate. And just to be sure we understand each other, that is and always will be using a 3% inflation rate 88.84% of the safe withdrawal rate after 5 years, 76.64% after 10 years and 57.04% after 20 years. In the words of the great Sir Jack, “Costs Matter”.
I'm not sure what you are accusing me of, but please do note that the title of my related post on this issue is "Retirement Income and the Tyranny of Compounding Fees" and it begins with a quote from John Bogle.
What I agreed about with Michael are his calculations for the impact on safe withdrawal rates, and not on the topic that fees hardly matter. They do matter a great deal.
At any rate, there are so many variables to deal with in retirement that it is nice to see how changing different variables impacts the same baseline case. That baseline case is constant inflation-adjusted spending. Yes, that is not an optimal strategy. Optimal strategies evolve with market performance. But we need some shared baseline to isolate the impacts of different assumptions.