Tuesday, April 15, 2014

When “Probably” Isn't Good Enough

Financial planners have long discussed whether we should pay off the mortgage, or keep a mortgage and invest that capital in the stock market.

The naive argument is that one can borrow a mortgage at say, 4%, and invest the money for a 10% return in the stock market, netting 6% on the “arbitrage”. But paying off a 4% mortgage is risk-free. (It is not a function of stock market returns.) The 10% market return in this example has a standard deviation of 20%, far from risk-free, based on historical market returns since 1926.

If you could find a risk-free 10% and keep the mortgage, that would be a certain 6% net and true arbitrage. But, that ain't gonna happen.

You would, in fact, earn the 6% difference in the years in which you receive the expected market return of 10%, but half of annual market returns are expected to be more than 10% and half less.

The meaning of “10% expected return with a standard deviation of 20%” is that in about two out of three years the returns will fall within the range of a 10% loss to a 30% gain (+ or - one standard deviation from the mean). That means the net return after the 4% mortgage payments will fall between a 16% loss and a 26% gain two out of three years. In 2007 to 2009, market returns fell way over to the wrong side of two-thirds of returns.

The worst financial mistake I have made in retirement (I should add “so far”) was taking the advice of a well-known financial planner in Washington before I retired in 2005. I could have bought my retirement home with cash, but he convinced me to hold a mortgage and leave that cash in the market. He gave me this advice because he thought the market would probably return about 10% a year on average over the life of the mortgage. But probably isn't the same as certainly and average isn't the same as annually.

Two years later, in 2007, the stock market crashed. Had I paid for the house with cash, I would have had hundreds of thousands fewer dollars exposed to the stock market.

The housing market crashed, too, but I don't plan on selling my home for a very long time, so that wasn't painful. Even the stock market crash was tolerable because I held a quite large helping of bonds. But it bugged the hell out of me that I had borrowed money against my home and paid a lot of interest for the privilege of losing that borrowed money in the stock market and that I was advised to do so by someone I trusted.

Fortunately, I could handle the losses. The former home of a good friend remains empty to this day after that housing and stock market crash. I see it every day, falling into disrepair. His mortgage was foreclosed, which brings up the most important reason I know to not borrow a mortgage to buy stocks. If you buy stocks on margin and the market crashes, you may have to sell some of your stocks and take a loss. If you buy stocks margined with your home and the market crashes, you can lose your home.

My friend had a lot of money going into 2007. It didn't seem at all likely that he would lose his home within a year.

Losing my home is high on my list of unacceptable outcomes.

I just found a 2011 article by Michael Kitces entitled Why Is It Risky To Buy Stocks On Margin But Prudent To Buy Them "On Mortgage"? This column and the comments adequately vet the issues, so I'm not going to repeat it or start a similar thread here. The issue I actually want to bring up is one of the magnitude and probability of risks.

The Retirement Income Industry Association  (RIAA) describes this in perhaps the most stilted prose you can encounter. I include it here for your literary amusement:
“The diverging opinions about the value of flooring in practice may derive from the range of opinions – that cannot be proven a priori – about a practitioner’s (or a client’s) view of the primacy of the probability of failure vs. the magnitude of failure. Some believe that the probability of failure looms greater in people's minds than the magnitude of failure. Others believe that consequences always trump the odds. There is no way to tell, a-priori, who is right and who is wrong for a specific set of circumstances whose resolution and outcomes are yet ahead of us.”
There. Got that?

What they're trying to say, sort of, is that at some point the consequences of failing are so horrible that “probably won't happen” is no longer good enough for some people. This is an ongoing struggle in financial planning with the “probabilities” group arguing that you can invest in stocks and you probably won't go broke and the “safety first” group countering that you should first make sure that nothing really bad is going to happen and invest what's left over in stocks.

Note that neither group says that you can't lose a whole lot of money. One side merely argues that you probably won't. The other side agrees but argues that “probably” isn't good enough when it comes to losing your standard of living in old age.

I lean toward the safety-first school. My tendency is to first take the unacceptable outcomes off the table and as I mentioned, I consider losing my home an unacceptable outcome.

I'm not against people taking financial risk in retirement. Given the difficulty involved in funding a retirement with our current system, there is no risk-free way to achieve it. I'm only against people taking risks they don't understand.

I've talked with many people who lost their homes or fortunes and not one was able to say, "I understood the risks when I took them and, if I had it to do over, I would make the same bet."

My former neighbor was a business school graduate and he could probably explain the details of his mortgage and his foreclosure risks. I'm just not sure he ever internalized that risk until it was too late.

My point isn't specifically about mortgages, or stock investments, or when to combine the two. It's about risk, both its probability and magnitude. Don't dismiss a potentially catastrophic outcome because the probability of it happening is very low.

You might go broke with a systematic withdrawals strategy, but you probably won't.

You might lose your home if you take a large mortgage and invest it in stocks, but you probably won't.

You could come up short if you claim Social Security benefits early and live to 90, but you probably won't.

The question you have to answer is when “probably won't” is good enough for you.


  1. I don't plan to have a mortgage in retirement. My wife and I do have a mortgage now even though we could sell investment assets and pay it off. However, I am not retired. "Probably won't" is good enough when a failure means that I simply have to keep working. (I am 37, nowhere near standard retirement age.)

    1. This comment has been removed by the author.

    2. I don't think it's the same scenario when you're still working.

      Before we retire, we take out a mortgage because most people can't afford to pay cash for a home. We are borrowing against future income (salary) we expect to earn. That makes perfect sense.

      After we retire, a mortgage is in part, and perhaps largely, borrowing against future expected stock market returns. Working longer is usually not an option. That's much riskier.

      At your age, it should be good enough. It changes pretty dramatically after you retire. Thanks for reading!

    3. Steve, we paid off our mortgage shortly before we retired for the reasons Dirk refers to in his article. At your age, you're doing great! You might find this article at FatWallet particularly interesting: "What Does an Extreme Early Retirement REALLY Require To Work?" Here's the URL: http://www.fatwallet.com/forums/finance/1357506/

  2. Dirk,

    Your made me smile.

    Yes, I wrote that in the Preface of the RMA Curriculum Book and, indeed, my Franglish can be weird.

    On the other hand, Precisely Stilted and Accurately Wise, what's not to love?

    This was the beginning of the articulation of Probability First vs. Safety First which is indeed a great literary improvement in the expression of it.



    1. English is a second language for me, as well -- I was born and raised in the South.

  3. Well, I am a new reader to this blog. I am 50 years old, a lawyer and starting to think about all of this stuff. What I am learning is that the thought process, the strategies and the risks are critically different between the "accumulation stage" and the "draw down".

    Mr Cotton, I think you bring up an important issue, and I whole heartedly agree. I always felt uncomfortable with planners telling me that statistically speaking, I should get X return with this or that allocation and that the Monte Carlo "simulation" gives me a 95% probability of success.

    Financial planning often looks backwards to foresee the future. Maybe that is because there is not another option. But it DOES NOT mean that (I) the particular statistic will be accurate or (ii) that, with a 95% success probability, you will not fall into the painful position of being in the unlucky 5%.

    So I fall into the safety first camp too. I like the retirement scenario that includes a guaranteed income "floor" (annuities, social security, cash) to cover all critical living costs plus a contingency.

    If you have net worth over and above that--- in my book, you can then consider investing in the market and playing the odds.

  4. If predicting the financial future from the past worked, it would be a lot easier, wouldn't it? Humans are incredibly bad at predicting the future, even a few years in advance. As people like Wade Pfau and Michael Kitces have shown, there is plenty of reason to believe the financial future won't look like the past and that "safe" withdrawal rates may be closer to 3% than 4%.

    I think there is a better way. I prefer scenario planning where you look at several possible futures, develop a plan that considers them all, and then consider the probabilities of each scenario developing.

    Glad to have you as a reader!

    1. One of the things (of several) that statistics assumes is that results of the sample are a reasonable approximation of the results of the population to which one wishes to generalize. I certainly am attuned to statistics (one of my favorite subjects in grad school), but I also learned in those classes that sometimes paying attention to individual cases can teach us a lot, as long as the "paying attention" is done systematically. So, JNEW, what really worked for me as I approached retirement was to record my personal, month-over-month expenses. Once I did that, and had an adequate sample of my own expenses (a couple of years, minimum), I then could generalize (predict) from MY history of past spending to MY probability of future spending.

      While I have some things in common with some populations, I also have some uniqueness that is best revealed in my personal financial data. You might give this strategy a try. It's certainly a cheap methodology!

  5. Hi Dirk, great blog. Like JNEW, I have also recently stated thinking about this stuff at age 52. If I'm understanding this post, your point is that you shouldn't have financed the home because the money you left in the market suffered significant losses in 2007-2009. But haven't you more than made up for it from 2009-2013? And today, aren't you in fact better off? I'm not sure if I'm missing something or what, but it seems that if someone has 7-10+ years before needing some portion of their money, the best place to keep that money is in stocks.

  6. I'm not better off.

    As you say, if someone has 7-10+ years before needing some of their money, the best place to keep that money is in stocks, in my opinion. But the situation is different once you retire and you are spending from that stash.

    I wrote about this in a blog post entitled "Even Your Portfolio Heals More Slowly as You Get Older". The "market" recovers much faster than does the portfolio of someone who is spending from their portfolio, as common sense would tell you. At 52, your portfolio probably recovered much faster than even the market because you weren't spending, but were, I assume, actually adding more savings.

    Because I am spending, and in fact spending more than I planned with kids still in college and med school, my portfolio has not yet recovered. It might never recover. Unplanned expenses happen.

    To exacerbate the market losses that I have not recovered, I subjected myself to foreclosure risk unnecessarily. I didn't need that extra leverage.

    Your question allows me to reiterate an important point about retirement: important things change after you retire. Many strategies that made sense before you retired don't make as much sense after.

    Thanks for writing! Your question is greatly appreciated.

  7. Your discussion about the 10 percent expected annual return, but with a standard deviation of plus or minus 20 percent, reminds me [don't worry, I'm not going to get into standard deviation type geek stuff here]:

    As an example, let's just simply say we have a 20 percent return this year but a 20 percent loss the subsequent year. Let's say we start with $100K, and let's say we're doing year-over-year losses, which is what annual expected returns generally amount to. We have first a 20 percent gain, bringing us to $120K, and then a loss of 20 percent times $120K, which is a loss of $24K, leaving us with only $96K.

    A 20 percent loss to start, followed by a 20 percent gain, has us traveling from $100K to $80K, then to the same $96K. It's sobering to contemplate that to break even after a 20 percent loss, one has to experience a 25 percent gain over the subsequent time period ($20K compared with $80K).

    My mathematical point is that an expected annual return, calculated over many time periods, appears to understate the actual return it it's considered on a year-over-year basis instead of on a starting point basis. That 10 percent expected annual return isn't as good as it appears -- especially if the standard deviation is plus or minus 20 percent [but I said I wasn't going to go there!], which suggests there'll be negatives of greater magnitude than just 20 percent.

    Mr. Cotton, you may well have pointed out this sort of thing in earlier blog posts, in which case, do feel free to delete my comment. I am a brand new reader here (and I'm very impressed by the soundness and simplicity [a good and desirable term, not at all a demeaning one] of your writing.