Thursday, January 3, 2019

Rational Expectations

It’s not easy to choose your portfolio’s asset allocation and decide how to change it as you head into retirement. William J. Bernstein discusses the many considerations that affect asset allocation decisions in his book Rational Expectations: Asset Allocation for Investing Adults, the last of four books in his Investing for Adults series.

Rather than just provide a set of model portfolios, the book includes a range of topics to help the reader choose his or her own allocation. These topics include historical asset-class returns, factor tilts, market efficiency, asset location, risk tolerance, your brain, liability matching portfolio, ETFs, rebalancing, and a lot more. You’re not left entirely on your own in building a portfolio; there are tables of recommended funds to cover the different asset classes.

To explain the intended meaning of “adult,” Bernstein says, “If the terms ‘standard deviation,’ ‘correlation,’ and ‘geometric average’ are Greek to you, or if you think you can time the market or pick stocks, then don’t buy this book.”

One quote that illustrates how Bernstein generally thinks clearly and correctly about investing is “The purpose of investing is not to simply optimize returns and make yourself rich. The purpose is to not die poor.” To this end, the author suggests that once you’ve saved enough to cover your future basic living expenses for the rest of your life, you should sell risky assets and buy government inflation-protected bonds. You can then invest any additional savings as riskily as you like.

Future stock returns are likely to be lower than they were in the past because it used to be difficult and expensive to own even a fraction of a stock index, “Prices stayed low, and the financial rewards were correspondingly high.” Further, “as societies grow wealthier, the supply and demand balance of capital shifts in favor of its consumers (companies) and away from its providers (investors).” The result is lower expected returns.

“Once you’ve arrived at a prudent asset allocation, tweaking it in one direction or the other makes relatively little difference to your long-term results.” What matters more is “your ability to stick to it through thick and thin.”

“When academics aggregate stock return data from many nations, statistically significant evidence of mean reversion appears. But since all the data come from the same time period, it’s hard to tell.” The existence of mean reversion means annual returns are correlated over time, which invalidates portfolio optimization methods that assume independence over time.

“Asset returns behave approximately lognormally,” but not exactly. A portfolio optimization technique called mean-variance optimization (MVO) assumes lognormal returns, and often gives crazy results. These crazy results come from a combination of the lognormal assumption and MVO being very sensitive to the accuracy of estimates of expected returns. Rather than use MVO, “I suggest a more profitable way to design your portfolio: stuff half the money in your mattress and lend the other half to your drug-addled nephew.”

On the subject of bonds, “sooner or later, we’re going to have an inflationary crisis, and in such an environment, long duration will be a killer.” “Bondholders of developed nations can expect to suffer permanent capital loss that equity holders will avoid.” For your fixed-income component, Bernstein recommends sticking to short-term government bonds and CDs (the U.S. equivalent of Canadian GICs). I don’t often find others who agree with the decision I made in my own portfolio to avoid corporate bonds and long-term government bonds.

Unfortunately, Bernstein has muddled ideas about asset location. He “largely” disagrees with the idea that “you should adjust downward the assets in your ... accounts by their estimated future tax exposures.” He reasons that assets are fungible, so “why should the exact stock/bond allocation of each [account] matter?”

The allocation in each account doesn’t matter. It’s the overall after-tax allocation that matters. By focusing on pre-tax asset allocation, Bernstein draws the incorrect conclusion that you should put assets that produce higher returns in accounts with lower expected tax rates. In reality, the author is mistaking tax efficiency for choosing a higher after-tax allocation to risky higher-return assets. When we hold the after-tax asset allocation constant, the supposed tax efficiency disappears.

The book quotes some interesting work by Philip Tetlock on market forecasters. The most extreme forecasts are the most wrong, but the media prefers extreme forecasts, and forecasters prefer media attention. “This sets up a positive feedback loop in which the media not only seeks out the most inaccurate forecasters but also worsens their forecasts.”

“Q. How do we know that economists have a sense of humor? A. They use decimal points.” It’s better to think clearly and make decisions that are approximately correct rather than rely too heavily on models that give precisely wrong results.

“Because threshold rebalancing tends to ‘catch’ market peaks and valleys more effectively than simple calendar rebalancing, I believe it is likely to be superior to calendar rebalancing.” I agree, but I wouldn’t recommend watching your portfolio daily in case a big market move triggers a need to rebalance. I’ve set up a spreadsheet that alerts me by email if I should rebalance my portfolio.

As for rebalancing thresholds, Bernstein gives some advice and concludes with “If you wind up rebalancing an asset class more than once per year on average, you should widen your thresholds. If you’re hardly rebalancing at all, narrow the thresholds.” The problem is that appropriate thresholds depend on portfolio size. My own work in this area may seem complex, but it factors in portfolio size, and in the end it’s just a few lines on a spreadsheet.

Overall, this is a valuable book for helping investors think about their asset allocations in sensible ways. I recommend it to anyone who meets the author’s definition of an “adult” investor.


  1. "I’ve set up a spreadsheet that alerts me by email if I should rebalance my portfolio."

    {Head explodes}

    I didn't know that was even possible. Now I'll have to Google how to do it (unless you're taking interns?)

    1. @Potato: I use a script that runs daily to check my Google Docs spreadsheet. The spreadsheet calculates cells that indicates whether rebalancing is necessary, and the script checks these cells.

      Here's the code I use with parts specific to me Xed out

      function PortfolioEmail() {
      var spreadsheet = SpreadsheetApp.openById("XXXXXX");
      var sheet = spreadsheet.getSheetByName("XXXXXX");
      // Fetch the range of cells B1..E1
      var dataRange = sheet.getRange(1, 2, 1, 4)
      // Fetch values for each row in the Range.
      var data = dataRange.getValues();
      for (i in data) {
      var row = data[i];
      var rebalance = row[0];
      var buy = row[2];
      var emailAddress = row[3];
      if (rebalance > 0 && buy > 0)
      var message = "Rebalance and Invest excess cash!";
      var subject = "Rebalance and Invest excess cash";
      MailApp.sendEmail(emailAddress, subject, message);
      if (rebalance > 0 && buy == 0)
      var message = "Rebalance";
      var subject = "Rebalance!";
      MailApp.sendEmail(emailAddress, subject, message);
      if (rebalance == 0 && buy > 0)
      var message = "Invest excess cash!";
      var subject = "Invest excess cash";
      MailApp.sendEmail(emailAddress, subject, message);

  2. I really liked all of Investing for Adults series. Could you explain why appropriate rebalancing thresholds depend on portfolio size? I've heard various authorities say that there is no evidence that any of the commonly used thresholds or time intervals is any better than another.

    1. @Grant: Rebalancing based on time is simpler, but using thresholds does a better job of capturing big moves in asset prices. One problem with thresholds is having to monitor, but I do that with a spreadsheet that emails me when necessary. Another problem is choosing window sizes. So, if you want 25% in foreign stocks, should your trigger for rebalancing be 20-30% or 24-26%. The answer is that sensible window widths depend on portfolio size. 20-30% is suitable for a small portfolio where transaction costs are significant. Large portfolios can use 24-26%. I give a lot more detail in a post I wrote:

  3. Ok, I see. The corollary to that is when you have a small allocation of an asset class regardless of the size of the portfolio, eg. 5% Reits. Larry Swedroe advises using either 5% or 25% rebalancing bands, whichever is the smaller. So with 5% Reits you would use 25% of the 5%, whereas with 50% bonds you would use 5% off target.

    1. @Grant: Swedroe's rule still leaves small portfolios trading too often creating high costs and leaves big portfolios trading too infrequently and leaving money on the table. It's certainly not among the top things to worry about in a portfolio, but optimal rebalancing bands depend on portfolio size.

    2. Thanks, Michael, I get it now!