In a sense, I added “researcher” to my résumé in the summer of 2005 when I attended the Wharton/IPI Private Wealth Management Program, a one-week program held twice a year at the Wharton Executive Education Center. During that week I suddenly realized that there was a gaping hole in the study of equity and asset allocation research: nobody had completed a study to determine the after-tax return on equity portfolios taxed at the top bracket.

Encouraged by Charlotte Beyer and intellectually stimulated by Professor Dick Marston, the founders of the program, I resolved to construct an unabridged model of the S&P 500 Index back to its founding in 1957 to determine what happened to the portfolios of those taxed at the then-prevailing top bracket in each of those years. The model also took into account the impact of state taxation. There had been, up to that point, attempts to study net returns but those studies did not specifically observe the impact on top bracket taxation; they omitted state taxation; and/or they did not examine post-liquidation results. Jean Brunel, editor of the Journal of Wealth Management, agreed to publish the resulting paper which was co-authored by Michael Blum. Since 2013, the model has been updated and refined by Scott B. Seibert, CFA. Scott’s work over the past five years has been instrumental not only in our after-tax research but in the formulation and publication of the Efficient Valuation Hypothesis revised paper.

The main objective of the paper was not only to calculate the after-tax return on equities, but also to compare stock and bond returns on a net basis. Thus, we were trying to observe the after-tax equity risk premium (or lack of risk premium) that was available to investors throughout the period. We supposed (when we began the study) that there were times when bonds outperformed equities on a net basis. That supposition turned out to be correct. When examining rolling 20-year periods beginning in 1957, we found that municipal bond portfolios outperformed equity portfolios 17% of the time. Just as important, we realized that this forward-looking outperformance could be observed at the beginning of the period in half of those instances.

Without the 2006 study and subsequent updates over the following decade, the Efficient Valuation Hypothesis (discussed in Chap. 7) might not have come into being. We believed then, and we staunchly believe now, that the notion that stocks outperform fixed income over time (popularly referred to as stocks for the long runFootnote 1) is false when examining net returns over specific periods. We proved in that paper that the equity risk premium (the excess return earned in a stock portfolio versus a bond portfolio) in the highest bracket is greatly diminished (or even negative at specific times) compared with a non-taxable portfolio such as an endowment or a qualified retirement account, which together provide the baseline used by most investors when making investment decisions. Thus, if an IRA investor invests in the S&P and earns 10%, a taxable investor would receive 7%, which is why high net worth investors need a more customized approach when making investment decisions.

When you review the tables in this chapter, you may find yourself over-emphasizing the averages that were observed over the entire length of the study. To avoid this, I recommend that you train your eyes on the shorter 20-year period to see how remarkably different the returns actually were.

To summarize:

  • The price (P/E ratio) one pays for a business or a group of businesses matters.

  • The tax rate, inclusive of taxes on dividends, long-term capital gains, and state taxes, matters.

  • The interest rate available on a 20-year non-taxable municipal bond in the investor’s home state matters.

Effect of a High Tax Bracket on After-Tax Returns on Stocks 1957 to 2017

We chose to begin our study of after-tax equity risk premium with 1957 because that is when the Standard & Poor’s 500 Index was first published.Footnote 2 We ended our study with December 31, 2017. Table 11.1 illustrates a hypothetical model portfolio, an investor in the highest tax bracket who invested $100 on January 1, 1957. The stocks in the portfolio match the performance of the S&P 500 Index inclusive of dividends. The model portfolio’s turnover rate is 5% for the base illustration. We also modeled for a 20% turnover rate, which we believe represents a “core” investment style for active managers. Note that the actual turnover rate of the S&P 500 is assumed to be 5% over the same period, and the turnover of the average actively managed mutual fund is assumed to be as high as 90%.

Table 11.1 S&P 500 taxed at top income bracket
Full size table

As we drilled down into more detail from the base model, we examined the variations in after-tax returns with turnover rates ranging from 0 to 100%. In each of the years, the portfolio paid taxes on dividends at the prevailing top tax rate, and all gains resulting from portfolio turnover were taxed at the prevailing long-term capital gains rates. The base model assumes a 6% state tax rate, but, as you will see, we also examined variations in state tax rates from 0 to 13%. As you will see, there is a remarkable variation in net returns earned on a portfolio owned by an investor domiciled in a no-tax state such as Florida and one living in a high-tax state such as California or New York.

The annualized, after-tax return for a high net worth investor who began investing in 1957 and who liquidated her portfolio on December 31, 2017, was 7.36%. To normalize for any long-term investment period since 1957, we analyzed each rolling 10- and 20-year periodFootnote 3 and concluded that the average annualized return for a 10-year period was 7.40% and 7.97% for a 20-year period. The median returns for these two investment horizons are 6.77% and 7.65%, respectively.Footnote 4

In the model, we simulated an actual portfolio with an initial $100 investment. The dividend yield during the first year was 1.75%. The S&P 500 dropped by 8.05% in 1957, and the annual dividend totaled $1.66, as a result of the declining value of the portfolio during the year. The tax on dividends in 1957 was 91% (the highest recorded taxation level), and we assumed a state tax of an additional 6%. After payment of federal and state taxes on dividends, the $1.66 was reduced to only $0.05. The portfolio turnover was assessed at a rate of 5%, which resulted in a capital gains tax loss carry-forward for the first year.

Impact of State of Residence on After-Tax Returns

The investor or trust’s state of residence has a notable effect on the long-term returns of equity portfolios because of variations in the rate of taxation on investment gains and dividends. In the above example, we modeled for a state income tax rate of 6% and concluded that the long-term compounded after-tax return of the equity portfolio was 7.36%. A portfolio that benefited from zero state tax in such states as Alaska, Florida, Nevada, South Dakota, Texas, Washington, or Wyoming would witness its annualized after-tax compounded return rise to 7.70%.

The zero-state-tax equity portfolio achieved an additional 34 basis points annualized return over the 6% state tax portfolio. Measured over many decades on a multimillion dollar portfolio, there is a significant capital accumulation advantage for the zero-state-tax portfolios.

As it happens, many portfolios of wealthy families are domiciled in high-income-tax states such as California with a 13.3% state tax, Vermont with an 8.95% state tax, Oregon with a 9.9% state tax, or the District of Columbia with an 8.95% tax. Investors living in New York City experience a combined city and state income tax rate of 12.7%. When we modeled the after-tax equity portfolio with a state tax rate of 13.3 (still using a turnover rate of 5%), the annualized return dropped to 6.92%. The difference between the zero-state-tax portfolios and the California portfolio was 78 basis points.

Impact of High Turnover Rates on Gross Returns

Portfolio turnover results in substantial costs to the investor. In building and studying the model, we observed portfolio turnover rates ranging from 0 to 100%. This allows those investors who prefer low-turnover equity portfolios to higher-turnover strategies to participate. Some investors opt for an equity mutual fund, and they will experience the fund’s average turnover, which can be as high as 90%. In our view, the 20% turnover rate we used as the secondary benchmark in our model was appropriate based on observations of large capitalization “core” portfolios. Note, as you examine Table 11.1, how much variation exists in the after-tax returns of portfolios that had identical gross returns yet achieved markedly lower net results. Equally notable is the effect, positive and negative, that state taxes had on net returns under varying portfolio turnover rates:

  • The most favorable return (8.28%) occurred in the portfolio with a 0% turnover rate domiciled in a 0% income tax state, and

  • The least favorable return (3.91%) came from the portfolio with a 100% portfolio turnover rate (366-day holding period in order to qualify for long-term capital gains tax rates) domiciled in a 13% income tax state. This fact might give pause to even the most talented day traders.

The difference between these two portfolios is 4.37%, which, over several decades, would have significant accumulation implications on a multimillion dollar portfolio. In dollar terms, if the two portfolios were to begin with $100 million, the one with no state tax and no turnover costs would earn an additional $11.8 billion over the 61 years we studied.

The Role of Time Horizon as a Predictor of Investment Results

As we studied the model and adjusted for state income taxes and portfolio turnover rates, the more we felt compelled to study shorter time horizons to see if it was possible for us to spot patterns, trends, valuation ratios, and characteristics that would have been visible to an investor before deploying capital in a portfolio. As a result of this analysis, we found convincing evidence that the most reliable time horizon for equity portfolios is 20 years because of the high correlation rate between earnings yield and rolling 20-year compounded performance. We also believe that 20-year observation periods give the investor a basis for comparing equities to the Bond Buyer 20 General Obligation Bond Index (which has an average maturity of 20 years). We discussed in detail in Chap. 7 the utility of using earnings yield in assessing the long-term return of equities. In observing the earnings yield of the portfolio in each year and the corresponding annualized returns the portfolio would achieve for each rolling 20-year period, we found a 0.79 correlation with the after-tax return the portfolio ultimately would achieve at the end of that same period.

The 10-year returns were more volatile than the 20-year periods, with a correlation rate of 0.61 between the earnings yield at the beginning of each 10-year rolling period and the after-tax return the portfolio would ultimately achieve at the end of that same period.

These data points prove that the effect of P/E ratio expansion and contraction has such a strong impact on actual returns over short time periods (such as 10 years) that simple valuation observations become less reliable predictive tools. As a result, it became clear that the earnings yield, with a few exceptions, was one of the best predictors of future returns on an after-tax basis (see Table 11.2).

Table 11.2 Twenty-year equity rolling returns (1957–2017)
Full size table

In examining 20-year periods we found:

  • The lowest after-tax return for equities of any 20-year period began on January 1, 1959, and ended on December 31, 1978. The annualized return for this period was 2.97%.

  • The highest after-tax return of any 20-year period began on January 1, 1980, and ended on December 31, 1999. The annualized return for this period was 14.75%.

  • The average after-tax return for all 20-year rolling periods was 7.97; the median was 7.65%.

We thought the period from 1989 to 2008 was particularly interesting. Despite the fact that there were three recessions (1991, 2000, and 2008) and two bear markets each representing a nearly 50% loss (2000 to 2002 and 2008 to 2009), the annualized return for a portfolio that began on January 1, 1989, and was liquidated on December 31, 2008, was 6.87%. When reviewing the 20-year returns, the bull market that ended around 2000 stands out because the after-tax returns for periods beginning between 1978 and 1982 all exceeded 12.00%.

Upon inspection of the wide variations in the results of various vintages, we asked ourselves, what statistics were available to investors at the beginning of these periods that would have given clues to the ultimate return that would be earned over the 20-year periods? Our objective in studying this history was to extract statistics that would have been available to investors that would have enabled them to make informed decisions about the future returns of their portfolios and whether stocks or bonds would outperform the other over various periods (we will discuss this question in detail later in the chapter). Future rates of inflation, GDP growth, earnings growth, and multiple expansions/contractions are not easily predicted by an investor beginning a 20-year investment portfolio. Another unknown is the ultimate tax rate that will be in effect for the next 20 years, so the investor must choose between modeling the current tax rate or an estimated tax rate. Investors in 2001 might rightfully have estimated that portfolio taxation rates would be lower in the future given the Bush administration’s emphasis on removing or reducing the double taxation of dividends. Similarly, an investor in the early 1960s would have been justified in modeling a lower forward-looking tax rate, given President Kennedy’s insistence that the 91% top rate of taxation was a drag on the economy and, as he neared the end of his first term, acceptance of the idea was gaining momentum. The earnings yield (trailing 12-month earnings/price) is the one statistic that might have given investors a clue as to the actual after-tax return of their portfolios at the time of investment.

The Role of Time Horizon on Decision to Liquidate a Portfolio

As important as the year in which investors begin investing in equities is the year in which they liquidate their investments. Using ten-year investment periods as an example, an investor who started on January 1, 1991, and liquidated on December 31, 2000, achieved a 13.55% return, whereas an investor who began on January 1, 1993, and liquidated on December 31, 2002, achieved a 6.99% return (nearly 50% lower). The two investors shared eight years in which they received the same returns, but during the two years they did not, 2001 and 2002, the S&P declined by 13.04% and 23.37%, respectively.

The temptation to liquidate a portfolio after sub-par returns can eliminate a large portion of the gains experienced in the previous years. For example, if the investor who began in 1993 had waited another two years, his annualized return after taxes would have increased to 8.55%. It must be noted that ordinary income taxes and capital gains taxes were reduced during year 12, which, of course, affected the final liquidation return; however, even without the tax change he still would have gained. I use this example to illustrate the fact that bear markets and recessions can inflict a bigger hit on 10-year than on 20-year holding periods. The 1983–2002 vintage portfolio managed to produce an after-tax return of 10.33% despite the fact that the last year of the portfolio witnessed the trough of a 50% bear market.

After-Tax Impact on the Decision to Invest in Bonds versus Equities

The after-tax equity risk premium over municipal bonds from 1957 to 2017 was 1.83%. Comparing the 20-year return data with the Bond Buyer Index (also known as the Bond Buyer Index of 20 Year General Obligation Bonds) yield over a similar time period, we found the average equity premium was 2.01% with a median of 1.38%. Of the 42 rolling 20-year periods we studied, there were seven in which bonds outperformed equities. The first occurred during the 20-year period beginning in 1959 and ending in 1978. The bond portfolio would have returned 3.40%, while the equity portfolio returned 2.97%. The largest period of bond outperformance was 1982 to 2001. During this period, the bond portfolio returned 13.36%, and the equity portfolio returned 12.55% for a total outperformance of 81 basis points.

The largest period of equity outperformance was 1979 to 1998. In that period the bond portfolio returned 6.58%, and the equity portfolio returned 14.42% for a total equity outperformance of 7.84%. Another interesting observation giving further evidence of the link between earnings yield and future returns, the earnings yield of the equity portfolio in 1979 was 12.05%. That earnings yield was a predictor of the 20-year after-tax performance that followed resulting in a difference of 240 basis points 12.05% earnings yield versus 14.45% actual performance.

In order to compare the annualized compounded after-tax returns of the equity portfolio, we decided to illustrate a best- and worst-case municipal bond portfolio. While the straight-line average of the Bond Buyer Index from 1957 to 2017 was 5.52% (straight-line average of bond buyer yields Table 11.3), we wanted to illustrate the effect that the timing of bond purchases would have had on a real-life portfolio. Therefore, we created 20 different municipal bond portfolios, each with different starting years to minimize the impact of the yields at the time a new bond was purchased. The best-case portfolio had a compounded after-tax return of 7.6% (meaning that the bonds outperformed equities by 1.36%) and the worst-case portfolio had a 4.5% compounded return (meaning that equities outperformed bonds by 1.74%).

Table 11.3 Bond buyer versus 20-year equity rolling returns
Full size table

We requested input on our study from the investor and investment management communities, and many suggested that we examine an actively managed municipal bond portfolio or mutual fund that could be compared to the simulated portfolios of rolling 20-year Bond Buyer 20 Index returns. This was a good suggestion, and we decided to analyze hypothetical returns from the T Rowe Price Tax Exempt Municipal Bond Fund (an open-ended mutual fund containing a portfolio of municipal bonds that began on October 26, 1976).

Using the Morningstar Advisor Workstation, we modeled a hypothetical investment in the fund, (ticker PRTAX) with no sales load and re-investment of all dividend and capital gains distributions. From the initial investment at the inception of the fund on October 26, 1976, through year end December 31, 2017, the annualized return was 6%. Note that although the T Rowe fund does not does not go back as far as 1957 (the inception date of the S&P 500 Index), the returns of this portfolio even inclusive of fees tend to rhyme with our observation of the average yield of the Bond Buyer Index, which was 6.05%. Thus, an actively managed strategy investing in municipal bonds managed to achieve the average yields of the index over the same period. We felt it was necessary to address this point as early critics of our paper felt that observing the average yield of the index did not account for the impact of bonds called prior to maturity or lengthening or shortening of maturities over time.

It should be noted that the starting yield to maturity of a 20-year bond can be achieved only if the bond is non-callable (therefore capable of experiencing 20 years of coupon income), if the re-invested coupon income achieves an identical yield as the starting yield, and if the bond experiences no default in terms of missed coupon payments or principal returned to the investor at maturity.

We found that while yield to maturity is one of the most important variables considered by investors considering purchase, it is rarely achieved in reality. Bond investors would find it impossible to achieve the exact same re-investment yield in each year of a 20-year bond due to the natural fluctuation of interest rates. In addition, most general obligation coupon bonds have some degree of callability. Thus, the only real-life way to achieve a known long-term compounding rate is to purchase a non-callable home state zero coupon bond. Zero coupon bonds are priced at a discount to maturity value and, when purchased and redeemed at par, will experience the exact annualized compounded yield to maturity that appears on the purchase confirmation. Figure 11.1 shows the dollar compounding of an equity portfolio matching the performance of the S&P 500 Index (pre-tax), an equity portfolio matching the S&P 500 Index (after subtracting prevailing dividend and capital gains taxes), and a tax-free municipal bond portfolio matching the yield of the Bond Buyer 20 General Obligation Index.

Fig. 11.1
figure 1

Growth of $100 (1957–2017)

Full size image

The 1957–2017 study produced many interesting results:

  • Twenty-year rolling returns are the most reliable measurement period for both stock and bond portfolios.

  • The average equity premium (on a 5% turnover portfolio) over tax-exempt municipal bonds was 2.01%.

  • Earnings yield provided a 0.79 correlation to the actual after-tax return earned by investors over 20-year periods.

  • Tax-exempt bond portfolios outperformed equity portfolios in 17% of the rolling 20-year periods (7 out of 42).

  • A general correlation was observed between tax-exempt bond yields and the after-tax returns of equity portfolios over 20-year rolling periods.

  • High-turnover portfolios matching the performance of the S&P 500 Index and domiciled in high-income-tax states have under-performed municipal bonds after taxes and fees.

As you can see from all this data, for the high net worth investor there is a massive disparity between pre-tax and after-tax returns, which has a cumulative impact on the dollar compounding of a portfolio over time. As an individual investor, it is important that you grasp the idea that for taxable investors the spread between equity and fixed income returns is narrower than the spread for non-taxable portfolios. Thus, you must (from the inception of your portfolio) measure the potential outperformance of a stock portfolio, estimate the loss of return as a result of taxes, and make a reasonable comparison between the forward-looking returns of various asset classes.