For any portfolio that's expected to increase over time, it's better to stay invested. That's why dollar-cost averaging a lump sum (a bonus, for example) into a portfolio rather than investing it all immediately is not a good decision. There are definitely not CAGR penalties for investing early and often or selling late and often. In hindsight, one can figure out the best days to buy and sell, but it's impossible to do so in advance. It's another example of market timing not being possible. I'll say it again: if one expects a portfolio to rise over time (and we all do, or we would not invest in it), it's mathematically optimum to keep dollars invested in said portfolio for the maximum time possible.sophie wrote: ↑Sat Jun 09, 2018 5:41 pmWe've had discussions on this in the past and definitely it's time for another one!
Starting with your example: Let's assume that you're just starting retirement, and your portfolio is exactly 25x annual expenses, i.e. expenses are 4% of the portfolio size. So your 2% assumes that you keep no more than 6 months expenses in cash. That seems a bit thin and I personally wouldn't be comfortable with that. This cash is not just for routine expenses, it's also for things like financing a new roof or an unforeseeable medical incident, and also to ride out market corrections. Most people probably head into retirement with at least 1 year in cash (4%), and many financial advisor/gurus recommend 5 years (20%). Notice that at the 5 year level, you are talking quite a significant slice of the portfolio.
Maybe it's mathematically better to stay invested at all times and sell as needed, but I kinda doubt it. I ran some simulations once, and there are definitely CAGR penalties for selling too often, or during a downturn. There's also the sleep at night factor. Skimping on cash works better if your portfolio is significantly more than 25x expenses, but if you're right at that 4% edge, I don't think you can afford it.
Regarding the roof-replacement example, the solution is simple: Withdraw the funds from the portfolio. This is one of the main purposes of low volatility portfolios, to support withdrawals with minimum sequence-of-return risk. Why would I unbalance my portfolio by taking my roof funds from one of the four asset classes? Does a roof replacement justify increasing one's average bond duration by withdrawing solely from cash? I'd simply withdraw the funds from the portfolio. I would probably use the withdrawal to balance the portfolio, drawing from the highest allocation. This is simply the inverse of buying the lagging asset when in accumulation mode. If the portfolio was already perfectly balanced, I'd sell equal amounts of each asset. And of course this is where a single fund or two-fund portfolio makes things even simpler.
I think a lot of this discussion stems from essential differences in how we view portfolios. For example, I see a lot of discussion on this board (not by you, Sophie), agonizing over the performance of one of the asset classes. If we all could step back and view our portfolios as a whole, much of this anxiety could be avoided. Also, while we might be able to mentally separate our 401k portfolios from our taxable portfolios, etc., our wealth doesn't care. It's going to grow/fluctuate based on our overall asset allocation. My goal has always been to stay as close to my target allocation across a variety of accounts. And I hope to continue that in the future years, especially if I'm in withdrawal mode.
In summary, I find large cash allocations unnecessary, and counterproductive, unless they are part of a portfolio structure.