Warren Buffett’s Berkshire Hathaway has delivered outstanding results for investors over time. This article analyses how the results might’ve been different if Buffett and Munger equally weighted each stock market investment, rather than concentrating most of their capital on a smaller number of high-conviction bets.
The methodology and idea for this were inspired by Robert Hagstrom’s fantastic book, “The Warren Buffett Portfolio”. In the book, Hagstrom performed the analysis for the years 1988 to 1997. In this article, the analysis was extrapolated to years post and prior, spanning from 1978 to 2021. The results, followed by the methodology and limitations are outlined below.
The results are as follows:
The graph above shows that the actual portfolio (i.e. the “concentrated” portfolio that Buffett actually constructed) outperformed the equally weighted portfolio in most five-year intervals.
The annualised return of Berkshire’s actual portfolio was 16.7% from 1978-2021 compared with 15.1% if each position was equally weighted. Note that the 16.7% is below Berkshire Hathaway’s book value annualised return due to the other investments Buffett has made over the years. This means that any investment return Berkshire achieved through options, warrants, float, leverage, private equity or fixed interest would not be reflected in the stock market portfolio returns.
These same annualised returns are shown above as multiples of the original investment.
The result over the 40+ year history is that Berkshire Hathaway’s stock portfolio would’ve turned $1 into $880 compared with the equally weighted portfolio, which would’ve grown the $1 into $486.
Clearly, the result is still fantastic even with equally weighting each position. This is no surprise since most of Buffett’s investments would’ve been successful.
Note that these returns likely understate the actual return achieved since dividends were excluded for most stocks.
Why did the equally weighted portfolio underperform Berkshire’s actual portfolio?
What this study shows is that concentrating capital on the few investments that are high-probability bets will likely outperform a portfolio that spreads the capital more thinly across lower-conviction ideas. The underperformance of the equally weighted portfolio was not a result of Buffett’s smaller positions being poor performers. Instead, it was because the handful of businesses Buffett invested big into, on average, performed very well.
Analysing each ten-year interval, we can see that the periods where Buffett’s concentrated portfolio outperformed the equally weighted alternative were when he invested a large amount of capital in one or two businesses which moved the needle immensely. We can also see that in the decade of underperformance, the concentration in Coca-Cola dragged down the overall returns of the portfolio.
|Interval||Berkshire Portfolio Annualised Return||Portfolio With Equally Weighted Positions Annualised Return||Influencial Investment||Portfolio weight range||Portfolio median weight||Annualised return|
|1978-1987||23.2%||16.1%||GEICO||21.4% – 50.9%||32.4%||33.6%|
|1988-1997||26.9%||25.3%||Coca-Cola||20.7% – 43.0%||38.2%||29.0%|
|1998-2007||5.5%||9.0%||Coca-Cola||17.4% – 43.3%||36.5%||-0.8%|
|2008-2021||13.5%||11.9%||Apple (investment made in 2016)||5.0% – 47.1%||27.7%||45.1%|
As we can see, in the three periods of outperformance, GEICO, Coca-Cola and Apple were weighted heavily and performed well above the portfolio average.
The lesson from this study is that generating outstanding investment results is more difficult if each investment is weighted equally regardless of conviction or investment potential. Though the difference was not huge on an annualised basis, Buffett’s stock market portfolio would’ve been worth nearly half what it currently is if he weighted every investment equally. Furthermore, note that the equally weighted portfolio is still very concentrated when compared to most mutual funds, often only spreading capital amongst less than 20 positions. Taking this study further and splitting Berkshire’s capital evenly across 50-100 stocks would very likely lead to even more underperformance below Berkshire’s actual portfolio. This isn’t to say that diversification is bad. Instead, the lesson is that merely diversifying for the sake of diversification, whilst knowingly investing less in investments with the largest probability of success and greatest margin of safety will likely lead to a worse result.
To view the research spreadsheet, click here.
- Berkshire Hathaway’s portfolio displayed in their investment letters was used for each year
- In the years when 13F filings were easily traceable, additional stocks were added and the “Number of Shares” and “Market Value” columns were determined from fintel.io. fintel.io was used for the years from 2013 to 2021. Dataroma.com was used for the years 2006-2012. Between 2006-2012, the “Market Value” column was approximated using the percentage weighting of the stock and the total “Portfolio Value” listed on Dataroma’s “history” tab. For example, Costco’s 2010 market value is equal to 0.60% x $52.6b = $315.6m. The number of shares was then calculated by dividing the market value by the stock price attained from Yahoo Finance.
- The market values were used to calculate the percentage weighting of each position
- The yearly returns were calculated in multiple ways:
- For stocks held in the previous year, the return is calculated by dividing their share prices year-over-year and subtracting 1. For stocks not held in the previous year, but purchased in the current year, the “cost” mentioned in Berkshire Hathaway’s annual report was used instead of last year’s stock price. For stocks not owned in the previous year and not in the annual report, their stock market return over that year was attained by comparing the “Adjusted Closed” price at the end of the year with the “Adjusted Close” price on the first trading day of the year, from Yahoo Finance. For stocks no longer traded, the returns were attained from other sources, which could include the company’s annual report for that year or other websites. These are detailed in the spreadsheet.
- For stocks no longer listed in the period 1988-1997, the return was taken from Robert Hagstrom’s book, “The Warren Buffett Portfolio”.
Assumptions and Limitations
- Since Berkshire reports at year-end, it is assumed that no additions or subtractions to the holdings were made over the past 12 months, which would not be true. However, many positions would’ve been fully held throughout the year and the impact is likely to be relatively insignificant.
- Not every stock owned by Berkshire was available to be included. Therefore, the returns are not completely accurate. However, in most cases these collectively made up less than 20% of the overall portfolio and would likely not meaningfully impact the results.
- Some stocks had to be excluded because their return for the particular year could not be found.
- Dividends were not included for every stock. Assuming the dividend yield would’ve been roughly the same across all positions, this would have an insignificant impact on the comparison between the concentrated and equally weighted portfolio. However, the annualised return figures are understated due to this.