The Rational Walk
Intelligent Investing is not a "Random Walk"

Berkshire Arbitrage Opportunity Eliminated March 25, 2009


Over the past month, I wrote a few posts regarding potential arbitrage opportunities available due to discrepancies between the relative valuations of Berkshire Hathaway’s Class A and Class B shares.  The arbitrage opportunity identified in these posts has now been entirely eliminated as a more “normal” relationship between the Class A and Class B shares returned.  Let’s revisit what took place over the past month and the implications for the notion that the market for Berkshire Hathaway shares is “efficient”.

Relationship Between Class A and Class B

As I wrote on February 20, each Class B share carries 1/30 of the economic rights and 1/200 of the voting rights of a single Class A share.  During normal times, the Class B shares trade at roughly 1/30 of the price of a Class A share reflecting the difference in economic rights.  It is logical for the Class B shares to sometimes trade at a small discount relative to the 1:30 ratio given the lower voting rights.  However, in the case of Berkshire Hathaway, voting rights are not as relevant given Warren Buffett’s ownership interests as well as the legions of loyal shareholders who will always vote with Buffett which gives him effective majority control.  All of the details related to the differences in Class A and Class B shares are listed in this memo written by Warren Buffett.

Record Spread Develops on February 20

Based on my research, on February 20, a record high spread developed between Class A and Class B shares.  On that day at the close of trading, it was possible to purchase 32.26 B shares for the same price as a single A share.  Any A shareholder was free to sell a single A share and purchase 32 B shares plus pocket the change represented by the fractional 0.26 B share.  By doing so, the A shareholder would effectively increase his economic interest in the company by 7.53% (including the retention of the cash equivalent of the fractional share).  Granted, the A shareholder would now only have a fraction of his prior voting rights, but that appears to be the only downside, aside from potential tax implications related to the A sale which could be significant.  A long position would be maintained with a significant addition to the shareholder’s economic position.

It would also have been possible to make a move that would not bet on the direction of Berkshire’s stock price but only on the eventual narrowing of the historically wide A/B spread.  By shorting one A and purchasing 30 Bs, an investor could effectively bet on an eventual closing of the historic spread.  Regardless of the direction in which Berkshire shares trade, the investor could profit when the spread returns to more typical levels.  At that time, the A share would be repurchased with the proceeds of selling the 30 B shares.  The main risk here would be if the spread widens further and does not narrow again in the future.

Spread Persists But Narrows in Mid March

I revisited the situation on March 9 and noted that the spread had narrowed briefly but then became quite wide again.  On March 9, an investor could purchase 31.69 B shares for the same price as one A share.  During this timeframe, Berkshire continued flirting with multi-year lows.  On March 10, Class A shares advanced 15.9% and Class B shares advanced 19.3% which narrowed the spread significantly but did not eliminate it. 

Spread Eliminated on March 25

Over the past several days, the spread continued to decline gradually and it was entirely eliminated on March 25 when Class A shares closed at 86,850 and Class B shares closed at 2,900.   At these levels, the Class B shares actually traded at a small premium over the typical 1:30 ratio.  The chart below shows the behavior of the A/B ratio between my March 9 post and March 25:


What Does This Prove?

The point of this post is not to claim any special knowledge of the absolute price movements in Berkshire Hathaway shares.  On February 20, I had no idea whether Berkshire Hathaway shares would trade higher or lower over the next month.  However, anyone could see that the relative valuation between Class A and Class B was not behaving normally. 

It was possible to take advantage of this in at least two ways:

Scenario 1:  On February 20, any existing A shareholder could have sold his A share and used the proceeds to purchase 32 B shares and kept the cash equivalent of 0.29 B shares for reinvestment elsewhere.  He would have sacrificed some voting rights, but this is not material for nearly all shareholders.  Today, that same shareholder could have reversed this transaction and sold his 32 B shares and purchased one A (or alternatively, he could have sold 30 Bs and purchased 1 A, keeping the extra 2 B shares).  This investor would have had a long position in Berkshire throughout this timeframe. 

Scenario 2:  On February 20, an investor could have sold short one A share and used the proceeds to purchase 32 B shares.  Today, that investor could have sold 30 B shares and covered the short of the one A share.  The investor could either keep the 2 additional B shares or sell them.  In either case, he would have profited from the narrowing of the spread.  Note that the success or failure of this operation would not have depended on the direction of Berkshire’s share price movement but only on the narrowing of the spread between A and B shares.

Is the Market for Berkshire Shares “Efficient”

It is very difficult to make the case that the market for Berkshire Class A and Class B shares is “efficient” as defined by the efficient market hypothesis when an investor could exploit conditions like what I have described here.  There were numerous other investors pointing out the situation in February yet the wide spread persisted for months.  If an efficient market is supposed to reflect all public information in the pricing of securities, surely this market did not display efficiency over the past month. 

It is fair to ask whether I personally traded on this arbitrage opportunity over the past month.  Unfortunately, I did not act quickly enough.  All of my holdings in Berkshire Hathaway are B shares purchased in small quantities over a period of close to a decade.  I did not have any A shares to sell in exchange for Bs as noted in Scenario 1.  I also have not made it a practice to engage in short selling of any kind in the past and did not have a brokerage account set up to execute Scenario 2.  While I do not believe in un-hedged short selling for many reasons, I plan to set up my brokerage account to permit short selling for arbitrage opportunities like this in the future.

Berkshire Arbitrage Opportunities Revisited March 9, 2009

A/B Spread: Feb 20 to March 9, 2009

Last month, I wrote about potential opportunities to profit from pricing discrepancies between Berkshire Hathaway Class A and Class B shares.  While the arbitrage opportunity was reduced significantly by February 26, it has re-emerged again more recently.  If someone holding A shares decided to sell one A at the close this afternoon, he could have purchased 31.69 B Shares.  This is short of the record high figure of 32.26 on February 20, but still a historically wide spread.  The chart below shows the behavior of the A/B Ratio over the past three weeks. 

A/B Spread: Feb 20 to March 9, 2009

Click this link for closing prices for each class of stock during this 12 trading day period.

As I wrote last month, each Class B share has economic interests equivalent to 1/30 of each Class A share and voting interests equivalent to 1/200 of each Class A share.  While the power of Class B shareholders from a voting perspective is clearly inferior, most Berkshire shareholders would have no influence over corporate policy anyway given Warren Buffett’s personal holdings as well as shareholders who are very loyal to him.  Therefore, it seems logical that the B shares should trade very close to 1/30 of the A shares.

What could be causing the apparent arbitrage opportunity?  On February 23, I speculated that it could be due to buying interest from someone who did in fact have an interest in accumulating voting control.  This would have supported a theory that Berkshire was buying back stock.  Berkshire’s annual report released on February 28 would have revealed such a move, so this theory is most likely incorrect.

One other potential theory is that the B shares have been under pressure due to forced sales by the charities that hold shares donated by Warren Buffett.  The International Herald Tribune reported that the Gates Foundation has sold 36,056 Class B shares so far in 2009 to comply with tax laws.  When Warren Buffett made his pledge to the Gates Foundation (as well as family related foundations) in 2006, he pledged Class B shares.  Buffett would convert the Class A shares he owns into Class B shares prior to making the gifts, presumably to retain as much voting control over Berkshire as possible even while giving away a large percentage of his economic interest over time.  It is certainly possible that forced selling by the foundations could be depressing the Class B share price compared to where it would normally trade relative to Class A.

Berkshire B Class Discount Persists February 23, 2009

On Friday, February 20, I wrote about the unusually high discount on Berkshire Hathaway B Class shares relative to A Class shares.  It appears that the discount persisted on Monday, February 23 as well but slightly narrowed to just over 32 Bs per 1 A share given the A share closing price of $75,600 and the B share closing price of $2,360. 

There are many theories floating around regarding why this strange anomaly could exist.  After all, any A shareholder could convert each share they own into 32 Bs and end up with a much higher economic interest (albeit with less voting control).  Most ordinary investors (and even some institutions) would logically not place a high premium on voting power considering the fact that it would be very difficult to exert influence given Warren Buffett’s ownership and large following of loyal shareholders. 

If enough A shareholders sell their shares and simultaneously purchase Bs, one would think that the selling pressure would reduce the price of A shares and increase the price of B shares enough to eliminate the arbitrage opportunity.  Why is this not happening?

A participant on a message board that I frequently read pointed out an interesting set of observations posted on the Motley Fool.  I found his theory interesting because it could indicate higher than normal buying demand for the As by someone who cares enough about the higher voting power to pay a significant premium.  Who might care so much about voting control to, in effect, pay a 6% premium?  Could this be a signal that Berkshire is buying back stock?  I suspect we will know the answer on Saturday when the annual report is released.  Stay tuned.

Berkshire Hathaway A/B Arbitrage February 20, 2009

Berkshire A/B Spread: 2/2008 to 2/2009

Berkshire Hathaway traded in an extreme range on Friday, February 20, 2009 and closed at multi-year lows. I happen to consider Berkshire Hathaway to be severely undervalued and I will have more to say about valuation when the 2008 report comes out next week.  In the meantime, it is interesting to make a few observations regarding a remarkable arbitrage situation that persisted for most of the day and a record high closing discount on 30 B shares relative to one A.

Arbitrage Opportunities:  Rare in Efficient Markets

One of the main principles of efficient markets is that it should be very difficult or impossible to take advantage of arbitrage situations.  Essentially this means that it should be very difficult to profit from price differentials between substantially identical securities.  With Berkshire Hathaway, we have something of a case study on this topic given that there are two series of shares that trade every day.  Berkshire Hathaway A shares are the original shares of the company while Berkshire Hathaway B shares were created in 1996 in an effort to thwart unit trusts that were planning to purchase the high priced A shares and offer slices to small investors (at a heavy fee, of course). 

Berkshire A and B Class Shares:  Substantially Identical Securities

Berkshire issued B shares with 1/30 of the economic rights of an A share and 1/200 of the voting rights.  The details are outlined in this memo.  Any A share may be converted to 30 B shares at any time at the option of the A shareholder.  The reverse is not the case:  one cannot exchange 30 B shares for one A.  

While 30 B shares are not precisely identical to 1 A share, they are substantially identical for any small shareholder who is unlikely to exert an influence from a voting perspective.  Still, there should logically be some small premium for A shares and it is reasonable to think that 30 Bs should sell at a minor discount to 1 A.

Arbitrage and the A/B Ratio:  Significant Spreads since October 2008

Given that the A and B shares have been trading now for well over a decade, it is interesting to look at trends in the ratio over certain time intervals and to see how the A/B ratio behaves during periods of market volatility and general irrational behavior on the part of market participants.  The following chart displays the A/B trend over the past year. As noted before, in theory, 30 Bs should trade at the level of 1 A, or at a slight discount:

Berkshire A/B Spread: 2/2008 to 2/2009

Smaller Spread During Less Volatile Times

Taking the long view, one can see that normally, the spread is quite narrow, meaning that the A/B ratio is usually very close to 30.  In other words, most of the time, there seems to be little in the way of arbitrage opportunities given that the ratio often stays at the 30:1 level for months and even years at a time.

Berkshire A/B Spread - 2000 to 2009

But Wait … We’ve Seen This Movie Before …

Looking at the one year chart of the A/B ratio from 2000, it is clear that scenarios similar to what we have seen over the past five months have taken place in the past.  The common thread seems to be that periods of high volatility in the markets and in Berkshire stock seem to be accompanied by seemingly irrational movements in the A shares relative to the B shares. 

Berkshire A/B Spread:  2000

Well, What’s the Point?

Given that the ratio as of today’s close was a record high at 32.26 B shares per 1 A, it seems prudent to look at the types of scenarios that have caused wide spreads in the past.  Consider that any A shareholder could have sold his or her A share and purchased 32 B shares and kept the change.  This is remarkable because 32 Bs have a significantly higher economic value than 1 A.  Although voting rights are substantially higher for the As, realistically, who is in a position to exert any influence over a company dominated by Warren Buffett and legions of loyal shareholders? 

I suppose the larger point is why such anomalies could in fact exist for any length of time in an efficient market.  Could it be that emotions were running so high today that irrational selling pressure on the Bs relative to the As caused the large spread? 

For the underlying data used in this analysis (courtesy of Yahoo Finance), please click on this link.