Embedded Files
Early 2014:
The illustrations immediately below are all of the original BRK annual report data (all units for change are in percent). The orange shaded region in the left illustration represents the reported underperformance years. The last illustration (showing difference in now relative performance) is similar to the center one, except that the latter half of the data is colored red instead of blue. Notice on this last chart the meaningfully lower performance is in this latter half of the data (that shown in red), across all S&P performance. Please consider all of three of these charts collectively.
The illustrations immediately below are similar to the ordered ones above, except we substitute market value for book value. The orange shaded region again represents underperformance, here using the new metric. Here too notice on this last chart the meaningfully higher performance in the green data versus red performance data, across all S&P performance (and with the S&P being less volatile). Again, one should consider all of these charts in concert.
For those somehow still having difficulty with the first chart immediately above, there are 15 underperforming years. Below we show the pattern in the frequency only. For relative underperformance severity, we also need to consider the red data in the two rightmost figures, of both chart rows above.
4 years of underperformance for 1st third of tenure
5 years of underperformance for 2nd third of tenure
6 years of underperformance for 3rd third of tenure
One can also download the underlying raw file, from the bottom of this page to customize the analysis, including differences, and tail calculations. Exceptional talent can see their outperformance at some point regress (due to size, market efficiencies, or other reasons), and that's what we are seeing here. Outside the mania surrounding this topic, it will be important later to patiently look back on these moments and assess whether Buffett's devotees were right in any knee-jerk rebuttals that he is performing "as good as ever". Noted in twice published probability analysis in early 2014, as the coming years progress, we'll more clearly see the distinct measure in how BRK performs relative to the S&P.
Update in early 2015:
Given some media requests, we have gladly uploaded the new, updated performance data. It should be noted off the start that it's been an extraordinary 50-year run, which we know from the "The forever elusive α" article that well less than 1% of us will ever be able to touch. The additional information below simply satisfies requests for educational data/material on the technical nature of this outperformance across time. This one chart is presented here, which stringently focuses on the differential in BRK's performance (on Buffett's new MV methodology as well as the original BV basis; both in continuous percent). Similar to the earlier paragraphs, the orange years continue to reflect underperformance on either a BV or MV basis, though not both. And the darker orange years reflect joint underperformance on both BV and MV bases. Notice where the non-shaded regions are on the time series. As we substantiated last year by way of careful probability models, the shaded regions jammed into the recent decade(s) is certainly not merely a brief, streak of bad luck.
Starting with 2009's performance, Buffett began robustly and consistently noting the new MV standard versus the traditional BV comparison that had been part of his firm's hallmark principles. It's for this reason only, that we show the descriptive statistics for how BRK's performance measures up across different exclusive regimes. Note that the dispersion calculations for the recent 6 years is simply to be complete in the illustration, and also the BRK summary has compounded returns over 50 years that differ by within 0.1% due to slight rounding errors if one utilizes from their main disaggregated data. Even slight adjustments here and there, regarding the middle window of the trisected time series, won't change the basic observations.
The overall story still doesn't change at all from the analysis last year, where we already critically examined these characteristics before it finally became fixed into the recent 2014 annual letter. Though in the fresh chart immediately above one can more easily see how -in the most recent decade- there has been a greater reduction in overall relative performance versus the S&P using even the MV basis (Buffett says this is due to a size disadvantage). The fact that the frequency of underperformance occurrences, using this MV basis, is shown to be improving in the past half-dozen years is simply probabilistic noise we can be accustomed to with BRK's change to MV. For example, up until recently every two years we could have expected that the BV measure to outperform (either or both years) the vast majority of the time, but now that probability is near half the time. But switching to the more chaotic MV measure can allow us to obtain some level of outperformance (again in either or both years) back up to the vast majority of the time. 2014 and since, in that sense, fits our forward expectation made last year. Stepping back from these probability calculations, one can also see this in looking at how the BV and MV statistics have both changed and interacted, in just the past 6 years (BV has actually edged out MV by 1% in the past 25 years).
It is worth highlighting that now those who are earlier in their careers face larger inequality (implying ownership in a smaller net share of the investing market). The best probability research we have that comes close to this topic of extreme statistics is in the "Top 1%, across states" article, and it shows that a good amount of this inequality is due to large and growing populations. When Buffett's BRK career started, the U.S. population was 40% less than it is now. Putting this all together, we can hypothesize (in order to refute) that if size was the main hindrance in outperforming the markets, then now we would be witnessing a progressively growing batch of stock pickers able to take advantage of this proportionally smaller share of the market.
Leaving both BRK and the MV assessment aside, this frequency count phenomenon is a fascinating probability topic with deep applications across many fields; it will be the subject of a future theoretical article, likely on the Statistical Ideas site. Part of Buffett's bright legacy is the enthusiasm to always keep learning (which started well before BRK and BPL); and in that spirit hope you join us as we examine other probability and statistics topics.
A quick update that Pulitzer-winner Steve Lohr enjoyed the NYT article: "a smart look at Warren Buffett's track record, by the numbers" and additional recaps here.
Page updated
Google Sites
Report abuse