While we don’t believe that anyone can reliably predict short-term market returns or inflation, and we would never use such forecasts to drive our investment decisions, good financial planning requires us to project our clients’ net worth over time[1], and that requires us to make assumptions about long-term investment returns and inflation.  Because our clients rely on our projections to determine when they will reach financial independence and how near-term decisions might affect that timeline, these underlying assumptions really matter.

There are many methodologies a financial planner could use to derive these assumptions – some require more analytical rigor and some less.  We try to use the most academically-supported, market-based methodology for estimating each of these key variables, rather than rely on our own guesswork.  Our process is as follows:

• For expected stock returns, we use the Fama-French Three-Factor Model.  We start with the risk-free rate[2], add the equity risk premium[3], then add the long-term premiums we expect from tilting the portfolio towards small cap and value stocks.  From that, we subtract the fees charged by the mutual fund and ETF companies (typically Vanguard and Dimensional) through which we own these stock portfolios.[4]
• For expected bond returns, we use a similar building-block approach.  Bond risk comes primarily from two factors: term risk and credit risk.  These risk factors are captured in the duration and credit rating of each bond.  So, we start with the risk-free rate[5] and we add the term and credit risk premiums associated with the bond funds used in client portfolios[6], subtracting the fees charged by the fund companies.[7]
• For inflation, we use the market’s expectation, which we define as the difference in yield between 20-year nominal U.S. Treasury bonds and 20-year inflation-protected U.S. Treasury bonds (TIPS).[8]

Since all of these variables are dynamic and change with market conditions, we recalculate them quarterly, take the average of expected returns over the past five years, and update the assumptions used in our projections.

The expected return of each client’s portfolio is a function of their current asset allocation, i.e., their split between stocks and bonds.[9]  Because we manage most client portfolios using a “glidepath,” in which the portfolio becomes more conservative (with higher bond allocations and lower expected returns) over time, we model expected returns accordingly, decreasing portfolio expected returns with age.

Since estimated returns for both stocks and bonds are based on interest rates, and because interest rates are at historical lows, our current expected return assumptions are also quite low.[10]  One could argue that it is overly conservative to assume (as we do) that they will always stay that way, but 1) academic theory suggests that is the right way to do it, and 2) if interest rates rise, it is likely that inflation would as well, so real returns (which are all that matter) may not change that much.

The projections we build for clients – like all models – require us to make assumptions about the unknown.  Because our clients rely on these long-term models when making life and career decisions, it’s important that we apply analytical rigor to the process, and that requires a repeatable, evidence-based methodology for setting and updating all key inputs.

[1] We use financial planning software to build a comprehensive projection of each client’s wealth over time.  The fun part of that process is client-specific (i.e., modeling how a family’s income and expenses will change over time in a variety of different scenarios), but also important are the macro assumptions that are common across clients, with expected long-term investment returns and inflation most significantly impacting the projections.

[2] We use the nominal yield on the current 20-year U.S. Treasury bond as a proxy for the risk-free rate when calculating stock returns.

[3] We calculate the equity risk premium by measuring the difference between the annual returns of the S&P 500 index and the 10-year U.S. Treasury bond over the longest period of available data.  Professor Aswath Damodaran (NYU) maintains a database of this information dating back to 1928.

[4] As of June 15, 2021, the average fund-level expense ratio for our stock mutual funds and ETFs is 0.24%.

[5] We use the nominal yield on the current 3-month U.S. Treasury bill as a proxy for the risk-free rate for bonds.

[6] The mutual fund companies (i.e., Vanguard and Dimensional) report the term and credit risk characteristics of their funds each month.  The premiums associated with these characteristics can be calculated using historical data from the Federal Reserve.  Specifically, we use differences in historical yields between U.S. Treasuries with various durations and the 3-month U.S. Treasury bill to estimate the term premium and differences in historical yields between corporate bonds with various credit ratings and U.S. Treasuries with the same duration to estimate the credit risk premium.

[7] As of June 15, 2021, the average fund-level expense ratio for our bond mutual funds is 0.24%.

[8] The difference in these yields technically includes both an estimate of expected inflation as well as a premium for accepting the risk of unexpected inflation.  Because we assume that the entire spread is expected inflation, our assumption may be somewhat conservative (in that it modestly overstates expected inflation).

[9] Expected portfolio return = (% allocation to stocks * estimated stock return) + (% allocation to bonds * estimated bond return).  Geometric’s fees are also modeled in, but those are treated as an expense (i.e., coming out of cash flows) and not as a drag on portfolio return.

[10] As of March 31, 2021, our expected return assumption for stocks was 8.0%, the expected return for bonds was 1.8%, and expected inflation was 2.4%.