Monday, June 23, 2014

Buffett's Alpha

A group of statisticians has analyzed Warren Buffett's investing returns. By looking at Berkshire Hathaway's 13F filings, they estimate that Buffett's public stock investments have outperformed the market by a much narrower margin than Berkshire's growth in book value. They conclude that Buffett has enhanced his returns with cheap leverage, calculating that the effective interest rate on Berkshire's insurance float has historically been less than the T-bill rate.

In their own words:
"In summary, we find that Buffett has developed a unique access to leverage that he has invested in safe, high-quality, cheap stocks and that these key characteristics can largely explain his impressive performance."

Their study is imperfect because the 13F filings don't list all of Berkshire's investments. It's possible that Buffett's distressed-debt investments have earned much higher returns than his stock investments, in which case insurance float would have contributed less to his returns than the paper suggests. (On the other hand, the statisticians argue that Buffett's private investments have earned much lower returns than his stock investments.)

Nonetheless, it makes a compelling argument that Buffett's has increased his returns with a uncommon form of leverage. This goes against the typical narrative of Buffett's success. For instance, in the introduction to Buffett: The Making of an American Capitalist, Roger Lowenstein writes that "Most of what Buffett did was imitable by the average person" and "As an investor, Buffett eschewed the use of leverage."

Most of what Buffett does may or may not be imitable, but the one thing that sets him apart from other skilled investors-- the cheap leverage-- definitely isn't. Without talented insurance executives to underwrite well, the cost of Berkshire's float would not have been so low. Without Ajit Jain and his predecessors, Buffett would not be Buffett. Prem Watsa at Fairfax Financial tried to copy Buffett's strategy of investing insurance float, and Fairfax nearly blew up even though Watsa's returns have been comparable to Buffett's.

Contrary to the perception that Buffett is the greatest investor alive, his unlevered returns are good but no higher than those of many other people. It's his business model that's exceptional. Investors should keep that in mind whenever Buffett opines about valuations, the economy, or the likely level of future stock-market returns. His opinions about these things aren't what made him uniquely rich, so they aren't uniquely valuable.


  1. Good post, not sure whether it is entirely true as buffet had some famous market calls (think about the closing of his partnership at the beginning of the sixties). These decisions alone would be sufficient to outperform the market (with the rest being average. Even charlie munger once said that over the past forty years they probably had only 15 outstanding investments, with the rest being average, but sufficient to produce outstanding returns. But even if you admit that he had outstanding performance, your conclusion still holds, as the median performance of his recomendations could well be below average, whereas said positive outliers lead to above average performance...

    1. I didn't mean to suggest that Buffett is a below-average investor. He's very good at picking stocks, but I think there are other people who are equally good, like Peter Lynch and Stanley Druckenmiller.

      My argument is that ultra-cheap leverage from Berkshire's insurance float has increased Berkshire's book value growth by several percent annually. That extra few percent is the reason why Buffett is so rich. Without the leverage, he would probably be worth only $5-10 billion instead of $60 billion.

    2. Great site, love the writing & approach. If you read the Snowball, Buffett learned from some guy (I forget the name) that buying in own undervalued stock was great way to increase stock price. Likewise selling overvalued stock. He did that through the insurance subs & buying in shares of companies that owned Berkshire shares (like Blue Chip Stamps and Wesco). An elaborate structure of shell companies within shell companies owning shares of the parent and so on was created in order to obfuscate what was really going on.

    3. "Great site, love the writing & approach."

      Thank you. I haven't read The Snowball, so I appreciate the color.

  2. The folks at AQR do put out some thought provoking and interesting pieces from time to time. In this case, I struggled with a few bits of vagueness in the writeup, though obviously the takeaway that cheap leverage in the form of float has juiced Berkshire’s returns stands. (A second level thought is that some of the AQR principals may want to create or acquire an insurance company…)

    A couple thoughts / questions:

    1. I didn’t quite understand what the authors meant by “overall stock market”—I have a sense that this may mean more than just the S&P 500, though I didn’t see that spelled out anywhere.

    2. The authors would speak of average annual returns and volatility. I believe the former relates to compound annual rates of return (or geometric mean, not arithmetic), but if so then why does volatility matter? I know that Beta in practice is a deeply disappointing metric that doesn’t really work. The basis for Beta, at least partially makes sense – you shouldn’t be earning excess returns for risks / uncertainties / volatilities that are easy to diversify away. Yet time and again the authors refer to Sharpe Ratios and basically ignore Betas. Is this because Buffett himself was not diversified and hence they think a Sharpe Ratio makes more sense here? Or is it because they think a rational investor in Berkshire decades ago would be unlikely to be diversified, or…. Did I miss something on why Sharpe Ratios are useful and Beta is not? (I am familiar with the argument for why they are both useful or both useless but not that Sharpe is useful while Beta is not.) Is it because most funds don’t have materially long lockups and hence cannot survive bad results even during a boom? I suspect this is why, though I find it unsatisfactory given how rich and independent minded Buffett was even in 1970 or 1980 – he didn’t need outside investors or really the corporate form of Berkshire.

    1. You're right, it's weird that they never define "overall stock market." The appendix says they used data from CRSP, so my guess is that they compared him to the CRSP US Total Market Index.

      Re: creating an insurance company, probably easier said than done:

  3. I am not sure how to reconcile this paper on Buffett's Alpha with this one:

    The Alpha paper might be trying to prove that Buffett is nothing extraordinary - he essentially levered up and his investments went up, translating into an insane amount of compounding.

    The second paper claims that the portfolio beat the market on its own.

    I am afraid that one ( or both) studies ended up using data to back up a pre-conceived conviction they already had in the first place.


Note: Only a member of this blog may post a comment.